[AArch64][SME2] Add SME2 MLA/MLS builtins. (#75584)

Add SME2 MLA/MLS builtins.

8 files changed, 4048 insertions, 0 deletions

diff --git a/clang/include/clang/Basic/arm_sme.td b/clang/include/clang/Basic/arm_sme.td
index ce99ca82c1d3..2a9bc6870bf7 100644
--- a/clang/include/clang/Basic/arm_sme.td
+++ b/clang/include/clang/Basic/arm_sme.td
@@ -353,6 +353,219 @@ let TargetGuard = "sme2" in {
   def SVBMOPS : Inst<"svbmops_za32[_{d}]_m", "viPPdd", "iUi", MergeNone, "aarch64_sme_bmops_za32", [IsSharedZA, IsStreaming], [ImmCheck<0, ImmCheck0_3>]>;
 }
 
+// FMLA/FMLS
+let TargetGuard = "sme2" in {
+  def SVMLA_MULTI_VG1x2_F32 : Inst<"svmla_za32[_{d}]_vg1x2", "vm22", "f", MergeNone, "aarch64_sme_fmla_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLA_MULTI_VG1x4_F32 : Inst<"svmla_za32[_{d}]_vg1x4", "vm44", "f", MergeNone, "aarch64_sme_fmla_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLS_MULTI_VG1x2_F32 : Inst<"svmls_za32[_{d}]_vg1x2", "vm22", "f", MergeNone, "aarch64_sme_fmls_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLS_MULTI_VG1x4_F32 : Inst<"svmls_za32[_{d}]_vg1x4", "vm44", "f", MergeNone, "aarch64_sme_fmls_vg1x4", [IsStreaming, IsSharedZA], []>;
+
+  def SVMLA_SINGLE_VG1x2_F32 : Inst<"svmla[_single]_za32[_{d}]_vg1x2", "vm2d", "f", MergeNone, "aarch64_sme_fmla_single_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLA_SINGLE_VG1x4_F32 : Inst<"svmla[_single]_za32[_{d}]_vg1x4", "vm4d", "f", MergeNone, "aarch64_sme_fmla_single_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLS_SINGLE_VG1x2_F32 : Inst<"svmls[_single]_za32[_{d}]_vg1x2", "vm2d", "f", MergeNone, "aarch64_sme_fmls_single_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLS_SINGLE_VG1x4_F32 : Inst<"svmls[_single]_za32[_{d}]_vg1x4", "vm4d", "f", MergeNone, "aarch64_sme_fmls_single_vg1x4", [IsStreaming, IsSharedZA], []>;
+
+  def SVMLA_LANE_VG1x2_F32 : Inst<"svmla_lane_za32[_{d}]_vg1x2", "vm2di", "f", MergeNone, "aarch64_sme_fmla_lane_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVMLA_LANE_VG1x4_F32 : Inst<"svmla_lane_za32[_{d}]_vg1x4", "vm4di", "f", MergeNone, "aarch64_sme_fmla_lane_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVMLS_LANE_VG1x2_F32 : Inst<"svmls_lane_za32[_{d}]_vg1x2", "vm2di", "f", MergeNone, "aarch64_sme_fmls_lane_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+  def SVMLS_LANE_VG1x4_F32 : Inst<"svmls_lane_za32[_{d}]_vg1x4", "vm4di", "f", MergeNone, "aarch64_sme_fmls_lane_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_3>]>;
+}
+
+let TargetGuard = "sme2,sme-f64f64" in {
+  def SVMLA_MULTI_VG1x2_F64 : Inst<"svmla_za64[_{d}]_vg1x2", "vm22", "d", MergeNone, "aarch64_sme_fmla_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLA_MULTI_VG1x4_F64 : Inst<"svmla_za64[_{d}]_vg1x4", "vm44", "d", MergeNone, "aarch64_sme_fmla_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLS_MULTI_VG1x2_F64 : Inst<"svmls_za64[_{d}]_vg1x2", "vm22", "d", MergeNone, "aarch64_sme_fmls_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLS_MULTI_VG1x4_F64 : Inst<"svmls_za64[_{d}]_vg1x4", "vm44", "d", MergeNone, "aarch64_sme_fmls_vg1x4", [IsStreaming, IsSharedZA], []>;
+
+  def SVMLA_SINGLE_VG1x2_F64 : Inst<"svmla[_single]_za64[_{d}]_vg1x2", "vm2d", "d", MergeNone, "aarch64_sme_fmla_single_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLA_SINGLE_VG1x4_F64 : Inst<"svmla[_single]_za64[_{d}]_vg1x4", "vm4d", "d", MergeNone, "aarch64_sme_fmla_single_vg1x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLS_SINGLE_VG1x2_F64 : Inst<"svmls[_single]_za64[_{d}]_vg1x2", "vm2d", "d", MergeNone, "aarch64_sme_fmls_single_vg1x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLS_SINGLE_VG1x4_F64 : Inst<"svmls[_single]_za64[_{d}]_vg1x4", "vm4d", "d", MergeNone, "aarch64_sme_fmls_single_vg1x4", [IsStreaming, IsSharedZA], []>;
+
+  def SVMLA_LANE_VG1x2_F64 : Inst<"svmla_lane_za64[_{d}]_vg1x2", "vm2di", "d", MergeNone, "aarch64_sme_fmla_lane_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
+  def SVMLA_LANE_VG1x4_F64 : Inst<"svmla_lane_za64[_{d}]_vg1x4", "vm4di", "d", MergeNone, "aarch64_sme_fmla_lane_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
+  def SVMLS_LANE_VG1x2_F64 : Inst<"svmls_lane_za64[_{d}]_vg1x2", "vm2di", "d", MergeNone, "aarch64_sme_fmls_lane_vg1x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
+  def SVMLS_LANE_VG1x4_F64 : Inst<"svmls_lane_za64[_{d}]_vg1x4", "vm4di", "d", MergeNone, "aarch64_sme_fmls_lane_vg1x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_1>]>;
+}
+
+// FMLAL/FMLSL/UMLAL/SMLAL
+// SMLALL/UMLALL/USMLALL/SUMLALL
+let TargetGuard = "sme2" in {
+  // MULTI MLAL
+  def SVMLAL_MULTI_VG2x2_F16 : Inst<"svmla_za32[_{d}]_vg2x2", "vm22", "bh", MergeNone, "aarch64_sme_fmlal_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG2x4_F16 : Inst<"svmla_za32[_{d}]_vg2x4", "vm44", "bh", MergeNone, "aarch64_sme_fmlal_vg2x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG2x2_S16 : Inst<"svmla_za32[_{d}]_vg2x2", "vm22", "s", MergeNone, "aarch64_sme_smlal_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG2x4_S16 : Inst<"svmla_za32[_{d}]_vg2x4", "vm44", "s", MergeNone, "aarch64_sme_smlal_vg2x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG2x2_U16 : Inst<"svmla_za32[_{d}]_vg2x2", "vm22", "Us", MergeNone, "aarch64_sme_umlal_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG2x4_U16 : Inst<"svmla_za32[_{d}]_vg2x4", "vm44", "Us", MergeNone, "aarch64_sme_umlal_vg2x4", [IsStreaming, IsSharedZA], []>;
+
+  def SVMLAL_MULTI_VG4x2_S8 : Inst<"svmla_za32[_{d}]_vg4x2", "vm22", "c", MergeNone, "aarch64_sme_smla_za32_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG4x2_U8 : Inst<"svmla_za32[_{d}]_vg4x2", "vm22", "Uc", MergeNone, "aarch64_sme_umla_za32_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG4x4_S8 : Inst<"svmla_za32[_{d}]_vg4x4", "vm44", "c", MergeNone, "aarch64_sme_smla_za32_vg4x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG4x4_U8 : Inst<"svmla_za32[_{d}]_vg4x4", "vm44", "Uc", MergeNone, "aarch64_sme_umla_za32_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // MULTI MLSL
+  def SVMLSL_MULTI_VG2x2_F16 : Inst<"svmls_za32[_{d}]_vg2x2", "vm22", "bh", MergeNone, "aarch64_sme_fmlsl_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG2x4_F16 : Inst<"svmls_za32[_{d}]_vg2x4", "vm44", "bh", MergeNone, "aarch64_sme_fmlsl_vg2x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG2x2_S16 : Inst<"svmls_za32[_{d}]_vg2x2", "vm22", "s", MergeNone, "aarch64_sme_smlsl_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG2x4_S16 : Inst<"svmls_za32[_{d}]_vg2x4", "vm44", "s", MergeNone, "aarch64_sme_smlsl_vg2x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG2x2_U16 : Inst<"svmls_za32[_{d}]_vg2x2", "vm22", "Us", MergeNone, "aarch64_sme_umlsl_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG2x4_U16 : Inst<"svmls_za32[_{d}]_vg2x4", "vm44", "Us", MergeNone, "aarch64_sme_umlsl_vg2x4", [IsStreaming, IsSharedZA], []>;
+
+  def SVMLSL_MULTI_VG4x2_S8 : Inst<"svmls_za32[_{d}]_vg4x2", "vm22", "c", MergeNone, "aarch64_sme_smls_za32_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG4x2_U8 : Inst<"svmls_za32[_{d}]_vg4x2", "vm22", "Uc", MergeNone, "aarch64_sme_umls_za32_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG4x4_S8 : Inst<"svmls_za32[_{d}]_vg4x4", "vm44", "c", MergeNone, "aarch64_sme_smls_za32_vg4x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG4x4_U8 : Inst<"svmls_za32[_{d}]_vg4x4", "vm44", "Uc", MergeNone, "aarch64_sme_umls_za32_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // SINGLE MLAL
+  def SVMLAL_SINGLE_VG2x1_F16 : Inst<"svmla_za32[_{d}]_vg2x1",          "vmdd", "bh", MergeNone, "aarch64_sme_fmlal_single_vg2x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG2x2_F16 : Inst<"svmla[_single]_za32[_{d}]_vg2x2", "vm2d", "bh", MergeNone, "aarch64_sme_fmlal_single_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG2x4_F16 : Inst<"svmla[_single]_za32[_{d}]_vg2x4", "vm4d", "bh", MergeNone, "aarch64_sme_fmlal_single_vg2x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG2x1_S16 : Inst<"svmla_za32[_{d}]_vg2x1",          "vmdd", "s", MergeNone, "aarch64_sme_smlal_single_vg2x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG2x2_S16 : Inst<"svmla[_single]_za32[_{d}]_vg2x2", "vm2d", "s", MergeNone, "aarch64_sme_smlal_single_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG2x4_S16 : Inst<"svmla[_single]_za32[_{d}]_vg2x4", "vm4d", "s", MergeNone, "aarch64_sme_smlal_single_vg2x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG2x1_U16 : Inst<"svmla_za32[_{d}]_vg2x1",          "vmdd", "Us", MergeNone, "aarch64_sme_umlal_single_vg2x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG2x2_U16 : Inst<"svmla[_single]_za32[_{d}]_vg2x2", "vm2d", "Us", MergeNone, "aarch64_sme_umlal_single_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG2x4_U16 : Inst<"svmla[_single]_za32[_{d}]_vg2x4", "vm4d", "Us", MergeNone, "aarch64_sme_umlal_single_vg2x4", [IsStreaming, IsSharedZA], []>;
+
+  def SVMLAL_SINGLE_VG4x1_S8  : Inst<"svmla_za32[_{d}]_vg4x1",          "vmdd", "c", MergeNone, "aarch64_sme_smla_za32_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x1_U8  : Inst<"svmla_za32[_{d}]_vg4x1",          "vmdd", "Uc", MergeNone, "aarch64_sme_umla_za32_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x2_S8  : Inst<"svmla[_single]_za32[_{d}]_vg4x2", "vm2d", "c", MergeNone, "aarch64_sme_smla_za32_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x2_U8  : Inst<"svmla[_single]_za32[_{d}]_vg4x2", "vm2d", "Uc", MergeNone, "aarch64_sme_umla_za32_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x4_S8  : Inst<"svmla[_single]_za32[_{d}]_vg4x4", "vm4d", "c", MergeNone, "aarch64_sme_smla_za32_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x4_U8  : Inst<"svmla[_single]_za32[_{d}]_vg4x4", "vm4d", "Uc", MergeNone, "aarch64_sme_umla_za32_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // SINGLE MLSL
+  def SVMLSL_SINGLE_VG2x1_F16 : Inst<"svmls_za32[_{d}]_vg2x1",          "vmdd", "bh", MergeNone, "aarch64_sme_fmlsl_single_vg2x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG2x2_F16 : Inst<"svmls[_single]_za32[_{d}]_vg2x2", "vm2d", "bh", MergeNone, "aarch64_sme_fmlsl_single_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG2x4_F16 : Inst<"svmls[_single]_za32[_{d}]_vg2x4", "vm4d", "bh", MergeNone, "aarch64_sme_fmlsl_single_vg2x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG2x1_S16 : Inst<"svmls_za32[_{d}]_vg2x1",          "vmdd", "s", MergeNone, "aarch64_sme_smlsl_single_vg2x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG2x2_S16 : Inst<"svmls[_single]_za32[_{d}]_vg2x2", "vm2d", "s", MergeNone, "aarch64_sme_smlsl_single_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG2x4_S16 : Inst<"svmls[_single]_za32[_{d}]_vg2x4", "vm4d", "s", MergeNone, "aarch64_sme_smlsl_single_vg2x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG2x1_U16 : Inst<"svmls_za32[_{d}]_vg2x1",          "vmdd", "Us", MergeNone, "aarch64_sme_umlsl_single_vg2x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG2x2_U16 : Inst<"svmls[_single]_za32[_{d}]_vg2x2", "vm2d", "Us", MergeNone, "aarch64_sme_umlsl_single_vg2x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG2x4_U16 : Inst<"svmls[_single]_za32[_{d}]_vg2x4", "vm4d", "Us", MergeNone, "aarch64_sme_umlsl_single_vg2x4", [IsStreaming, IsSharedZA], []>;
+
+  def SVMLSL_SINGLE_VG4x1_S8  : Inst<"svmls_za32[_{d}]_vg4x1",          "vmdd", "c", MergeNone, "aarch64_sme_smls_za32_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x1_U8  : Inst<"svmls_za32[_{d}]_vg4x1",          "vmdd", "Uc", MergeNone, "aarch64_sme_umls_za32_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x2_S8  : Inst<"svmls[_single]_za32[_{d}]_vg4x2", "vm2d", "c", MergeNone, "aarch64_sme_smls_za32_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x2_U8  : Inst<"svmls[_single]_za32[_{d}]_vg4x2", "vm2d", "Uc", MergeNone, "aarch64_sme_umls_za32_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x4_S8  : Inst<"svmls[_single]_za32[_{d}]_vg4x4", "vm4d", "c", MergeNone, "aarch64_sme_smls_za32_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x4_U8  : Inst<"svmls[_single]_za32[_{d}]_vg4x4", "vm4d", "Uc", MergeNone, "aarch64_sme_umls_za32_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // INDEXED MLAL
+  def SVMLAL_LANE_VG2x1_F16 : Inst<"svmla_lane_za32[_{d}]_vg2x1", "vmddi", "bh", MergeNone, "aarch64_sme_fmlal_lane_vg2x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG2x2_F16 : Inst<"svmla_lane_za32[_{d}]_vg2x2", "vm2di", "bh", MergeNone, "aarch64_sme_fmlal_lane_vg2x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG2x4_F16 : Inst<"svmla_lane_za32[_{d}]_vg2x4", "vm4di", "bh", MergeNone, "aarch64_sme_fmlal_lane_vg2x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG2x1_S16 : Inst<"svmla_lane_za32[_{d}]_vg2x1", "vmddi", "s", MergeNone, "aarch64_sme_smlal_lane_vg2x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG2x2_S16 : Inst<"svmla_lane_za32[_{d}]_vg2x2", "vm2di", "s", MergeNone, "aarch64_sme_smlal_lane_vg2x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG2x4_S16 : Inst<"svmla_lane_za32[_{d}]_vg2x4", "vm4di", "s", MergeNone, "aarch64_sme_smlal_lane_vg2x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG2x1_U16 : Inst<"svmla_lane_za32[_{d}]_vg2x1", "vmddi", "Us", MergeNone, "aarch64_sme_umlal_lane_vg2x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG2x2_U16 : Inst<"svmla_lane_za32[_{d}]_vg2x2", "vm2di", "Us", MergeNone, "aarch64_sme_umlal_lane_vg2x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG2x4_U16 : Inst<"svmla_lane_za32[_{d}]_vg2x4", "vm4di", "Us", MergeNone, "aarch64_sme_umlal_lane_vg2x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+
+  def SVMLAL_LANE_VG4x1_S8 : Inst<"svmla_lane_za32[_{d}]_vg4x1", "vmddi", "c", MergeNone, "aarch64_sme_smla_za32_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLAL_LANE_VG4x1_U8 : Inst<"svmla_lane_za32[_{d}]_vg4x1", "vmddi", "Uc", MergeNone, "aarch64_sme_umla_za32_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLAL_LANE_VG4x2_S8 : Inst<"svmla_lane_za32[_{d}]_vg4x2", "vm2di", "c", MergeNone, "aarch64_sme_smla_za32_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLAL_LANE_VG4x2_U8 : Inst<"svmla_lane_za32[_{d}]_vg4x2", "vm2di", "Uc", MergeNone, "aarch64_sme_umla_za32_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLAL_LANE_VG4x4_S8 : Inst<"svmla_lane_za32[_{d}]_vg4x4", "vm4di", "c", MergeNone, "aarch64_sme_smla_za32_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLAL_LANE_VG4x4_U8 : Inst<"svmla_lane_za32[_{d}]_vg4x4", "vm4di", "Uc", MergeNone, "aarch64_sme_umla_za32_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+
+  // INDEXED MLSL
+  def SVMLSL_LANE_VG2x1_F16 : Inst<"svmls_lane_za32[_{d}]_vg2x1", "vmddi", "bh", MergeNone, "aarch64_sme_fmlsl_lane_vg2x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG2x2_F16 : Inst<"svmls_lane_za32[_{d}]_vg2x2", "vm2di", "bh", MergeNone, "aarch64_sme_fmlsl_lane_vg2x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG2x4_F16 : Inst<"svmls_lane_za32[_{d}]_vg2x4", "vm4di", "bh", MergeNone, "aarch64_sme_fmlsl_lane_vg2x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG2x1_S16 : Inst<"svmls_lane_za32[_{d}]_vg2x1", "vmddi", "s", MergeNone, "aarch64_sme_smlsl_lane_vg2x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG2x2_S16 : Inst<"svmls_lane_za32[_{d}]_vg2x2", "vm2di", "s", MergeNone, "aarch64_sme_smlsl_lane_vg2x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG2x4_S16 : Inst<"svmls_lane_za32[_{d}]_vg2x4", "vm4di", "s", MergeNone, "aarch64_sme_smlsl_lane_vg2x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG2x1_U16 : Inst<"svmls_lane_za32[_{d}]_vg2x1", "vmddi", "Us", MergeNone, "aarch64_sme_umlsl_lane_vg2x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG2x2_U16 : Inst<"svmls_lane_za32[_{d}]_vg2x2", "vm2di", "Us", MergeNone, "aarch64_sme_umlsl_lane_vg2x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG2x4_U16 : Inst<"svmls_lane_za32[_{d}]_vg2x4", "vm4di", "Us", MergeNone, "aarch64_sme_umlsl_lane_vg2x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+
+  def SVMLSL_LANE_VG4x1_S8 : Inst<"svmls_lane_za32[_{d}]_vg4x1", "vmddi", "c", MergeNone, "aarch64_sme_smls_za32_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLSL_LANE_VG4x1_U8 : Inst<"svmls_lane_za32[_{d}]_vg4x1", "vmddi", "Uc", MergeNone, "aarch64_sme_umls_za32_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLSL_LANE_VG4x2_S8 : Inst<"svmls_lane_za32[_{d}]_vg4x2", "vm2di", "c", MergeNone, "aarch64_sme_smls_za32_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLSL_LANE_VG4x2_U8 : Inst<"svmls_lane_za32[_{d}]_vg4x2", "vm2di", "Uc", MergeNone, "aarch64_sme_umls_za32_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLSL_LANE_VG4x4_S8 : Inst<"svmls_lane_za32[_{d}]_vg4x4", "vm4di", "c", MergeNone, "aarch64_sme_smls_za32_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVMLSL_LANE_VG4x4_U8 : Inst<"svmls_lane_za32[_{d}]_vg4x4", "vm4di", "Uc", MergeNone, "aarch64_sme_umls_za32_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+
+  // SINGLE SUMLALL
+  // Single sumla maps to usmla, with zn & zm operands swapped
+  def SVSUMLALL_SINGLE_VG4x1 : Inst<"svsumla_za32[_{d}]_vg4x1",          "vmdu",   "c", MergeNone, "aarch64_sme_usmla_za32_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+
+  def SVSUMLALL_SINGLE_VG4x2 : Inst<"svsumla[_single]_za32[_{d}]_vg4x2", "vm2.du", "c", MergeNone, "aarch64_sme_sumla_za32_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVSUMLALL_SINGLE_VG4x4 : Inst<"svsumla[_single]_za32[_{d}]_vg4x4", "vm4.du", "c", MergeNone, "aarch64_sme_sumla_za32_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // Multi-multi sumla builtins are mapped to usmla, with zn & zm operands swapped
+  def SVSUMLALL_MULTI_VG4x2 : Inst<"svsumla_za32[_{d}]_vg4x2", "vm2.d2.u", "c", MergeNone, "aarch64_sme_usmla_za32_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVSUMLALL_MULTI_VG4x4 : Inst<"svsumla_za32[_{d}]_vg4x4", "vm4.d4.u", "c", MergeNone, "aarch64_sme_usmla_za32_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // INDEXED SUMLALL
+  def SVSUMLALL_LANE_VG4x1 : Inst<"svsumla_lane_za32[_{d}]_vg4x1", "vmdui", "c", MergeNone, "aarch64_sme_sumla_za32_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVSUMLALL_LANE_VG4x2 : Inst<"svsumla_lane_za32[_{d}]_vg4x2", "vm2ui", "c", MergeNone, "aarch64_sme_sumla_za32_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVSUMLALL_LANE_VG4x4 : Inst<"svsumla_lane_za32[_{d}]_vg4x4", "vm4ui", "c", MergeNone, "aarch64_sme_sumla_za32_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+
+  // SINGLE USMLALL
+  def SVUSMLALL_SINGLE_VG4x1 : Inst<"svusmla_za32[_{d}]_vg4x1",          "vmdx",   "Uc", MergeNone, "aarch64_sme_usmla_za32_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+  def SVUSMLALL_SINGLE_VG4x2 : Inst<"svusmla[_single]_za32[_{d}]_vg4x2", "vm2.dx", "Uc", MergeNone, "aarch64_sme_usmla_za32_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVUSMLALL_SINGLE_VG4x4 : Inst<"svusmla[_single]_za32[_{d}]_vg4x4", "vm4.dx", "Uc", MergeNone, "aarch64_sme_usmla_za32_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // MULTI USMLALL
+  def SVUSMLALL_MULTI_VG4x2 : Inst<"svusmla_za32[_{d}]_vg4x2", "vm2.d2.x", "Uc", MergeNone, "aarch64_sme_usmla_za32_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVUSMLALL_MULTI_VG4x4 : Inst<"svusmla_za32[_{d}]_vg4x4", "vm4.d4.x", "Uc", MergeNone, "aarch64_sme_usmla_za32_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // INDEXED USMLALL
+  def SVUSMLALL_LANE_VG4x1 : Inst<"svusmla_lane_za32[_{d}]_vg4x1", "vmdxi", "Uc", MergeNone, "aarch64_sme_usmla_za32_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVUSMLALL_LANE_VG4x2 : Inst<"svusmla_lane_za32[_{d}]_vg4x2", "vm2xi", "Uc", MergeNone, "aarch64_sme_usmla_za32_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+  def SVUSMLALL_LANE_VG4x4 : Inst<"svusmla_lane_za32[_{d}]_vg4x4", "vm4xi", "Uc", MergeNone, "aarch64_sme_usmla_za32_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_15>]>;
+}
+
+let TargetGuard = "sme2,sme-i16i64" in {
+  // MULTI MLAL
+  def SVMLAL_MULTI_VG4x2_S16 : Inst<"svmla_za64[_{d}]_vg4x2", "vm22", "s", MergeNone, "aarch64_sme_smla_za64_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG4x2_U16 : Inst<"svmla_za64[_{d}]_vg4x2", "vm22", "Us", MergeNone, "aarch64_sme_umla_za64_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG4x4_S16 : Inst<"svmla_za64[_{d}]_vg4x4", "vm44", "s", MergeNone, "aarch64_sme_smla_za64_vg4x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_MULTI_VG4x4_U16 : Inst<"svmla_za64[_{d}]_vg4x4", "vm44", "Us", MergeNone, "aarch64_sme_umla_za64_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // MULTI MLSL
+  def SVMLSL_MULTI_VG4x2_S16 : Inst<"svmls_za64[_{d}]_vg4x2", "vm22", "s", MergeNone, "aarch64_sme_smls_za64_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG4x2_U16 : Inst<"svmls_za64[_{d}]_vg4x2", "vm22", "Us", MergeNone, "aarch64_sme_umls_za64_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG4x4_S16 : Inst<"svmls_za64[_{d}]_vg4x4", "vm44", "s", MergeNone, "aarch64_sme_smls_za64_vg4x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_MULTI_VG4x4_U16 : Inst<"svmls_za64[_{d}]_vg4x4", "vm44", "Us", MergeNone, "aarch64_sme_umls_za64_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // SINGLE MLAL
+  def SVMLAL_SINGLE_VG4x1_S16 : Inst<"svmla_za64[_{d}]_vg4x1",          "vmdd", "s", MergeNone, "aarch64_sme_smla_za64_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x1_U16 : Inst<"svmla_za64[_{d}]_vg4x1",          "vmdd", "Us", MergeNone, "aarch64_sme_umla_za64_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x2_S16 : Inst<"svmla[_single]_za64[_{d}]_vg4x2", "vm2d", "s", MergeNone, "aarch64_sme_smla_za64_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x2_U16 : Inst<"svmla[_single]_za64[_{d}]_vg4x2", "vm2d", "Us", MergeNone, "aarch64_sme_umla_za64_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x4_S16 : Inst<"svmla[_single]_za64[_{d}]_vg4x4", "vm4d", "s", MergeNone, "aarch64_sme_smla_za64_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLAL_SINGLE_VG4x4_U16 : Inst<"svmla[_single]_za64[_{d}]_vg4x4", "vm4d", "Us", MergeNone, "aarch64_sme_umla_za64_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // SINGLE MLSL
+  def SVMLSL_SINGLE_VG4x1_S16 : Inst<"svmls_za64[_{d}]_vg4x1",          "vmdd", "s", MergeNone, "aarch64_sme_smls_za64_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x1_U16 : Inst<"svmls_za64[_{d}]_vg4x1",          "vmdd", "Us", MergeNone, "aarch64_sme_umls_za64_single_vg4x1", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x2_S16 : Inst<"svmls[_single]_za64[_{d}]_vg4x2", "vm2d", "s", MergeNone, "aarch64_sme_smls_za64_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x2_U16 : Inst<"svmls[_single]_za64[_{d}]_vg4x2", "vm2d", "Us", MergeNone, "aarch64_sme_umls_za64_single_vg4x2", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x4_S16 : Inst<"svmls[_single]_za64[_{d}]_vg4x4", "vm4d", "s", MergeNone, "aarch64_sme_smls_za64_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+  def SVMLSL_SINGLE_VG4x4_U16 : Inst<"svmls[_single]_za64[_{d}]_vg4x4", "vm4d", "Us", MergeNone, "aarch64_sme_umls_za64_single_vg4x4", [IsStreaming, IsSharedZA], []>;
+
+  // INDEXED MLAL
+  def SVMLAL_LANE_VG4x1_S16 : Inst<"svmla_lane_za64[_{d}]_vg4x1", "vmddi", "s", MergeNone, "aarch64_sme_smla_za64_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG4x1_U16 : Inst<"svmla_lane_za64[_{d}]_vg4x1", "vmddi", "Us", MergeNone, "aarch64_sme_umla_za64_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG4x2_S16 : Inst<"svmla_lane_za64[_{d}]_vg4x2", "vm2di", "s", MergeNone, "aarch64_sme_smla_za64_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG4x2_U16 : Inst<"svmla_lane_za64[_{d}]_vg4x2", "vm2di", "Us", MergeNone, "aarch64_sme_umla_za64_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG4x4_S16 : Inst<"svmla_lane_za64[_{d}]_vg4x4", "vm4di", "s", MergeNone, "aarch64_sme_smla_za64_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLAL_LANE_VG4x4_U16 : Inst<"svmla_lane_za64[_{d}]_vg4x4", "vm4di", "Us", MergeNone, "aarch64_sme_umla_za64_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+
+  // INDEXED MLSL
+  def SVMLSL_LANE_VG4x1_S16 : Inst<"svmls_lane_za64[_{d}]_vg4x1", "vmddi", "s", MergeNone, "aarch64_sme_smls_za64_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG4x1_U16 : Inst<"svmls_lane_za64[_{d}]_vg4x1", "vmddi", "Us", MergeNone, "aarch64_sme_umls_za64_lane_vg4x1", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG4x2_S16 : Inst<"svmls_lane_za64[_{d}]_vg4x2", "vm2di", "s", MergeNone, "aarch64_sme_smls_za64_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG4x2_U16 : Inst<"svmls_lane_za64[_{d}]_vg4x2", "vm2di", "Us", MergeNone, "aarch64_sme_umls_za64_lane_vg4x2", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG4x4_S16 : Inst<"svmls_lane_za64[_{d}]_vg4x4", "vm4di", "s", MergeNone, "aarch64_sme_smls_za64_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+  def SVMLSL_LANE_VG4x4_U16 : Inst<"svmls_lane_za64[_{d}]_vg4x4", "vm4di", "Us", MergeNone, "aarch64_sme_umls_za64_lane_vg4x4", [IsStreaming, IsSharedZA], [ImmCheck<3, ImmCheck0_7>]>;
+}
+
 //
 // Spill and fill of ZT0
 //
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index a29304c81928..7bc3b7594c8f 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -10318,6 +10318,28 @@ Value *CodeGenFunction::EmitAArch64SVEBuiltinExpr(unsigned BuiltinID,
   return nullptr;
 }
 
+static void swapCommutativeSMEOperands(unsigned BuiltinID,
+                                       SmallVectorImpl<Value *> &Ops) {
+  unsigned MultiVec;
+  switch (BuiltinID) {
+  default:
+    return;
+  case SME::BI__builtin_sme_svsumla_za32_s8_vg4x1:
+    MultiVec = 1;
+    break;
+  case SME::BI__builtin_sme_svsumla_za32_s8_vg4x2:
+    MultiVec = 2;
+    break;
+  case SME::BI__builtin_sme_svsumla_za32_s8_vg4x4:
+    MultiVec = 4;
+    break;
+  }
+
+  if (MultiVec > 0)
+    for (unsigned I = 0; I < MultiVec; ++I)
+      std::swap(Ops[I + 1], Ops[I + 1 + MultiVec]);
+}
+
 Value *CodeGenFunction::EmitAArch64SMEBuiltinExpr(unsigned BuiltinID,
                                                   const CallExpr *E) {
   auto *Builtin = findARMVectorIntrinsicInMap(AArch64SMEIntrinsicMap, BuiltinID,
@@ -10340,6 +10362,9 @@ Value *CodeGenFunction::EmitAArch64SMEBuiltinExpr(unsigned BuiltinID,
            BuiltinID == SME::BI__builtin_sme_svstr_za)
     return EmitSMELdrStr(TypeFlags, Ops, Builtin->LLVMIntrinsic);
 
+  // Handle builtins which require their multi-vector operands to be swapped
+  swapCommutativeSMEOperands(BuiltinID, Ops);
+
   // Should not happen!
   if (Builtin->LLVMIntrinsic == 0)
     return nullptr;
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mla.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mla.c
new file mode 100644
index 000000000000..f52edd9888da
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mla.c
@@ -0,0 +1,292 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED,A5) A1##A3##A5
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3,A4,A5) A1##A2##A3##A4##A5
+#endif
+
+//
+// Multi, multi
+// CHECK-LABEL: @test_svmla2_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZM]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla2_f32j13svfloat32x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZM]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla2_f32(uint32_t slice_base, svfloat32x2_t zn, svfloat32x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_za32,_f32,_vg1x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla4_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 4)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], <vscale x 4 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla4_f32j13svfloat32x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], <vscale x 4 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla4_f32(uint32_t slice_base, svfloat32x4_t zn, svfloat32x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_za32,_f32,_vg1x4,,)(slice_base, zn, zm);
+}
+
+//
+// Multi, single
+// CHECK-LABEL: @test_svmla_single2_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.single.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single2_f32j13svfloat32x2_tu13__SVFloat32_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.single.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single2_f32(uint32_t slice_base, svfloat32x2_t zn, svfloat32_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_single_za32,,_f32,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single4_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.single.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single4_f32j13svfloat32x4_tu13__SVFloat32_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.single.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single4_f32(uint32_t slice_base, svfloat32x4_t zn, svfloat32_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_single_za32,,_f32,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, indexed
+// CHECK-LABEL: @test_svmla_lane2_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.lane.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane2_f32j13svfloat32x2_tu13__SVFloat32_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.lane.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane2_f32(uint32_t slice_base, svfloat32x2_t zn, svfloat32_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_lane_za32,_f32,_vg1x2,,)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svmla_lane4_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.lane.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane4_f32j13svfloat32x4_tu13__SVFloat32_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.lane.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane4_f32(uint32_t slice_base, svfloat32x4_t zn, svfloat32_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_lane_za32,_f32,_vg1x4,,)(slice_base, zn, zm, 3);
+}
+
+//
+// Multi, multi
+// CHECK-LABEL: @test_svmla2_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZM]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla2_f64j13svfloat64x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZM]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla2_f64(uint32_t slice_base, svfloat64x2_t zn, svfloat64x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_za64,_f64,_vg1x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla4_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 2)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], <vscale x 2 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla4_f64j13svfloat64x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], <vscale x 2 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla4_f64(uint32_t slice_base, svfloat64x4_t zn, svfloat64x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_za64,_f64,_vg1x4,,)(slice_base, zn, zm);
+}
+
+//
+// Multi, single
+// CHECK-LABEL: @test_svmla_single2_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.single.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single2_f64j13svfloat64x2_tu13__SVFloat64_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.single.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single2_f64(uint32_t slice_base, svfloat64x2_t zn, svfloat64_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_single_za64,,_f64,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single4_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.single.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single4_f64j13svfloat64x4_tu13__SVFloat64_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.single.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single4_f64(uint32_t slice_base, svfloat64x4_t zn, svfloat64_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_single_za64,,_f64,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, indexed
+// CHECK-LABEL: @test_svmla_lane2_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.lane.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane2_f64j13svfloat64x2_tu13__SVFloat64_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.lane.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane2_f64(uint32_t slice_base, svfloat64x2_t zn, svfloat64_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_lane_za64,_f64,_vg1x2,,)(slice_base, zn, zm, 1);
+}
+
+// CHECK-LABEL: @test_svmla_lane4_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.lane.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane4_f64j13svfloat64x4_tu13__SVFloat64_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmla.lane.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane4_f64(uint32_t slice_base, svfloat64x4_t zn, svfloat64_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmla_lane_za64,_f64,_vg1x4,,)(slice_base, zn, zm, 1);
+}
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mlal.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mlal.c
new file mode 100644
index 000000000000..834ade753507
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mlal.c
@@ -0,0 +1,696 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED,A5) A1##A3##A5
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3,A4,A5) A1##A2##A3##A4##A5
+#endif
+
+//
+// Multi, multi
+// CHECK-LABEL: @test_svmla2_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla2_f16j13svfloat16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla2_f16(uint32_t slice_base, svfloat16x2_t zn, svfloat16x2_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_f16,_vg2x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla2_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z16test_svmla2_bf16j14svbfloat16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla2_bf16(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16x2_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_bf16,_vg2x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla2_u16j12svuint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16x2_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_u16,_vg2x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla2_s16j11svint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla2_s16(uint32_t slice_base, svint16x2_t zn, svint16x2_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_s16,_vg2x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla4_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], <vscale x 8 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla4_f16j13svfloat16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], <vscale x 8 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla4_f16(uint32_t slice_base, svfloat16x4_t zn, svfloat16x4_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_f16,_vg2x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla4_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z16test_svmla4_bf16j14svbfloat16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla4_bf16(uint32_t slice_base, svbfloat16x4_t zn, svbfloat16x4_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_bf16,_vg2x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla4_u16j12svuint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16x4_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_u16,_vg2x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmla4_s16j11svint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla4_s16(uint32_t slice_base, svint16x4_t zn, svint16x4_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_s16,_vg2x4,,)(slice_base, zn, zm);
+}
+
+//
+// Multi, single
+// CHECK-LABEL: @test_svmla_single1_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x1.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[ZN:%.*]], <vscale x 8 x half> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single1_f16ju13__SVFloat16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x1.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[ZN:%.*]], <vscale x 8 x half> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single1_f16(uint32_t slice_base, svfloat16_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_f16,_vg2x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single1_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x1.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[ZN:%.*]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmla_single1_bf16ju14__SVBfloat16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x1.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[ZN:%.*]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single1_bf16(uint32_t slice_base, svbfloat16_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_bf16,_vg2x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single1_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.single.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single1_u16ju12__SVUint16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.single.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single1_u16(uint32_t slice_base, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_u16,_vg2x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single1_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.single.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single1_s16ju11__SVInt16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.single.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single1_s16(uint32_t slice_base, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_za32,_s16,_vg2x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single2_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single2_f16j13svfloat16x2_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single2_f16(uint32_t slice_base, svfloat16x2_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_single_za32,,_f16,,_vg2x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single2_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmla_single2_bf16j14svbfloat16x2_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single2_bf16(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_single_za32,,_bf16,,_vg2x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.single.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.single.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_single_za32,,_u16,,_vg2x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.single.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.single.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_single_za32,,_s16,,_vg2x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single4_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single4_f16j13svfloat16x4_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single4_f16(uint32_t slice_base, svfloat16x4_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_single_za32,,_f16,,_vg2x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single4_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmla_single4_bf16j14svbfloat16x4_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.single.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single4_bf16(uint32_t slice_base, svbfloat16x4_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_single_za32,,_bf16,,_vg2x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.single.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.single.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_single_za32,,_u16,,_vg2x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.single.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmla_single4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.single.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_single_za32,,_s16,,_vg2x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, indexed
+//
+
+// CHECK-LABEL: @test_svmla_lane1_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x1.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[ZN:%.*]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane1_f16ju13__SVFloat16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x1.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[ZN:%.*]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane1_f16(uint32_t slice_base, svfloat16_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_f16,_vg2x1)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane1_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x1.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[ZN:%.*]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_svmla_lane1_bf16ju14__SVBfloat16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x1.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[ZN:%.*]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane1_bf16(uint32_t slice_base, svbfloat16_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_bf16,_vg2x1)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane1_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.lane.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane1_u16ju12__SVUint16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.lane.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane1_u16(uint32_t slice_base, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_u16,_vg2x1)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane1_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.lane.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane1_s16ju11__SVInt16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.lane.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane1_s16(uint32_t slice_base, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_s16,_vg2x1)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane2_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane2_f16j13svfloat16x2_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane2_f16(uint32_t slice_base, svfloat16x2_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_f16,_vg2x2)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane2_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_svmla_lane2_bf16j14svbfloat16x2_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane2_bf16(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_bf16,_vg2x2)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.lane.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.lane.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_u16,_vg2x2)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.lane.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.lane.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_s16,_vg2x2)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane4_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane4_f16j13svfloat16x4_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane4_f16(uint32_t slice_base, svfloat16x4_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_f16,_vg2x4)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane4_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_svmla_lane4_bf16j14svbfloat16x4_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlal.lane.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane4_bf16(uint32_t slice_base, svbfloat16x4_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_bf16,_vg2x4)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.lane.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlal.lane.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_u16,_vg2x4)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmla_lane4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.lane.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmla_lane4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlal.lane.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_lane4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmla_lane_za32,,,_s16,_vg2x4)(slice_base, zn, zm, 7);
+}
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mlall.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mlall.c
new file mode 100644
index 000000000000..fceb16a48260
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mlall.c
@@ -0,0 +1,1790 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -S -target-feature +sme-i16i64 -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-i16i64 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED,A5) A1##A3##A5
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3,A4,A5) A1##A2##A3##A4##A5
+#endif
+
+//
+// Single x 1
+//
+
+// MLAL
+
+// CHECK-LABEL: @test_svmla_single_x1_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmla_single_x1_s8ju10__SVInt8_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x1_s8(uint32_t slice_base, svint8_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za32,_s8,_vg4x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single_x1_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.single.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmla_single_x1_s16ju11__SVInt16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.single.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x1_s16(uint32_t slice_base, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za64,_s16,_vg4x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_uvmlal_single_x1_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_uvmlal_single_x1_u8ju11__SVUint8_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_uvmlal_single_x1_u8(uint32_t slice_base, svuint8_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za32,_u8,_vg4x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_uvmlal_single_x1_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.single.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_uvmlal_single_x1_u16ju12__SVUint16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.single.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_uvmlal_single_x1_u16(uint32_t slice_base, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za64,_u16,_vg4x1,,)(slice_base, zn, zm);
+}
+
+// MLSL
+
+// CHECK-LABEL: @test_svmls_single_x1_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmls_single_x1_s8ju10__SVInt8_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x1_s8(uint32_t slice_base, svint8_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za32,_s8,_vg4x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single_x1_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.single.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmls_single_x1_s16ju11__SVInt16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.single.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x1_s16(uint32_t slice_base, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za64,_s16,_vg4x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_uvmlsl_single_x1_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_uvmlsl_single_x1_u8ju11__SVUint8_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_uvmlsl_single_x1_u8(uint32_t slice_base, svuint8_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za32,_u8,_vg4x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_uvmlsl_single_x1_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.single.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_uvmlsl_single_x1_u16ju12__SVUint16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.single.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_uvmlsl_single_x1_u16(uint32_t slice_base, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za64,_u16,_vg4x1,,)(slice_base, zn, zm);
+}
+
+// SUMLALL
+
+// CHECK-LABEL: @test_sumlall_single_x1_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZM:%.*]], <vscale x 16 x i8> [[ZN:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_sumlall_single_x1_s8ju10__SVInt8_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZM:%.*]], <vscale x 16 x i8> [[ZN:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_sumlall_single_x1_s8(uint32_t slice_base, svint8_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svsumla_za32,_s8,_vg4x1,,)(slice_base, zn, zm);
+}
+
+// USMLALL
+
+// CHECK-LABEL: @test_usmlall_single_x1_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_usmlall_single_x1_u8ju11__SVUint8_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.single.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_usmlall_single_x1_u8(uint32_t slice_base, svuint8_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svusmla_za32,_u8,_vg4x1,,)(slice_base, zn, zm);
+}
+
+//
+// Single x 2
+//
+
+// MLAL
+
+// CHECK-LABEL: @test_svmla_single_x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmla_single_x2_s8j10svint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x2_s8(uint32_t slice_base, svint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_single_za32,,_s8,,_vg4x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single_x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.single.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmla_single_x2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.single.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_single_za64,,_s16,,_vg4x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single_x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmla_single_x2_u8j11svuint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x2_u8(uint32_t slice_base, svuint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_single_za32,,_u8,,_vg4x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single_x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.single.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmla_single_x2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.single.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_single_za64,,_u16,,_vg4x2)(slice_base, zn, zm);
+}
+
+// MLSL
+
+// CHECK-LABEL: @test_svmls_single_x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmls_single_x2_s8j10svint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x2_s8(uint32_t slice_base, svint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_single_za32,,_s8,,_vg4x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single_x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.single.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmls_single_x2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.single.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_single_za64,,_s16,,_vg4x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single_x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmls_single_x2_u8j11svuint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x2_u8(uint32_t slice_base, svuint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_single_za32,,_u8,,_vg4x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single_x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.single.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmls_single_x2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.single.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_single_za64,,_u16,,_vg4x2)(slice_base, zn, zm);
+}
+
+// SUMLALL
+
+// CHECK-LABEL: @test_svsumla_single_x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_svsumla_single_x2_s8j10svint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svsumla_single_x2_s8(uint32_t slice_base, svint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svsumla_single_za32,,_s8,,_vg4x2)(slice_base, zn, zm);
+}
+
+// USMLALL
+
+// CHECK-LABEL: @test_usmlall_single_x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_usmlall_single_x2_u8j11svuint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.single.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_usmlall_single_x2_u8(uint32_t slice_base, svuint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svusmla_single_za32,,_u8,,_vg4x2)(slice_base, zn, zm);
+}
+
+//
+// Single x 4
+//
+
+// MLAL
+
+// CHECK-LABEL: @test_svmla_single_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmla_single_x4_s8j10svint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x4_s8(uint32_t slice_base, svint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_single_za32,,_s8,,_vg4x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single_x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.single.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmla_single_x4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.single.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_single_za64,,_s16,,_vg4x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single_x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmla_single_x4_u8j11svuint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x4_u8(uint32_t slice_base, svuint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_single_za32,,_u8,,_vg4x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmla_single_x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.single.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmla_single_x4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.single.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmla_single_x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_single_za64,,_u16,,_vg4x4)(slice_base, zn, zm);
+}
+
+// MLSL
+
+// CHECK-LABEL: @test_svmls_single_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmls_single_x4_s8j10svint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x4_s8(uint32_t slice_base, svint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_single_za32,,_s8,,_vg4x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single_x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.single.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmls_single_x4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.single.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_single_za64,,_s16,,_vg4x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single_x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmls_single_x4_u8j11svuint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x4_u8(uint32_t slice_base, svuint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_single_za32,,_u8,,_vg4x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single_x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.single.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z24test_svmls_single_x4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.single.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single_x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_single_za64,,_u16,,_vg4x4)(slice_base, zn, zm);
+}
+
+// SUMLALL
+
+// CHECK-LABEL: @test_svsumla_single_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_svsumla_single_x4_s8j10svint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svsumla_single_x4_s8(uint32_t slice_base, svint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svsumla_single_za32,,_s8,,_vg4x4)(slice_base, zn, zm);
+}
+
+// USMLALL
+
+// CHECK-LABEL: @test_usmlall_single_x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_usmlall_single_x4_u8j11svuint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.single.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_usmlall_single_x4_u8(uint32_t slice_base, svuint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svusmla_single_za32,,_u8,,_vg4x4)(slice_base, zn, zm);
+}
+
+//
+// Multi x 2
+//
+
+// MLAL
+
+// CHECK-LABEL: @test_mlal_multi_x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_mlal_multi_x2_s8j10svint8x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlal_multi_x2_s8(uint32_t slice_base, svint8x2_t zn, svint8x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za32,_s8,_vg4x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlal_multi_x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_mlal_multi_x2_s16j11svint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlal_multi_x2_s16(uint32_t slice_base, svint16x2_t zn, svint16x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za64,_s16,_vg4x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlal_multi_x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_mlal_multi_x2_u8j11svuint8x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlal_multi_x2_u8(uint32_t slice_base, svuint8x2_t zn, svuint8x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za32,_u8,_vg4x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlal_multi_x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_mlal_multi_x2_u16j12svuint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlal_multi_x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za64,_u16,_vg4x2,,)(slice_base, zn, zm);
+}
+
+// MLSL
+
+// CHECK-LABEL: @test_mlsl_multi_x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_mlsl_multi_x2_s8j10svint8x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlsl_multi_x2_s8(uint32_t slice_base, svint8x2_t zn, svint8x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za32,_s8,_vg4x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlsl_multi_x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_mlsl_multi_x2_s16j11svint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlsl_multi_x2_s16(uint32_t slice_base, svint16x2_t zn, svint16x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za64,_s16,_vg4x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlsl_multi_x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_mlsl_multi_x2_u8j11svuint8x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlsl_multi_x2_u8(uint32_t slice_base, svuint8x2_t zn, svuint8x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za32,_u8,_vg4x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlsl_multi_x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_mlsl_multi_x2_u16j12svuint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlsl_multi_x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za64,_u16,_vg4x2,,)(slice_base, zn, zm);
+}
+
+// SUMLALL
+
+// CHECK-LABEL: @test_sumlal_multi_x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_sumlal_multi_x2_s8j10svint8x2_t11svuint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_sumlal_multi_x2_s8(uint32_t slice_base, svint8x2_t zn, svuint8x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svsumla_za32,_s8,_vg4x2,,)(slice_base, zn, zm);
+}
+
+// USMLALL
+
+// CHECK-LABEL: @test_usmlal_multi_x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_usmlal_multi_x2_u8j11svuint8x2_t10svint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_usmlal_multi_x2_u8(uint32_t slice_base, svuint8x2_t zn, svint8x2_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svusmla_za32,_u8,_vg4x2,,)(slice_base, zn, zm);
+}
+
+//
+// Multi x 4
+//
+
+// MLAL
+
+// CHECK-LABEL: @test_mlal_multi_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_mlal_multi_x4_s8j10svint8x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlal_multi_x4_s8(uint32_t slice_base, svint8x4_t zn, svint8x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za32,_s8,_vg4x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlal_multi_x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_mlal_multi_x4_s16j11svint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlal_multi_x4_s16(uint32_t slice_base, svint16x4_t zn, svint16x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za64,_s16,_vg4x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlal_multi_x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_mlal_multi_x4_u8j11svuint8x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlal_multi_x4_u8(uint32_t slice_base, svuint8x4_t zn, svuint8x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za32,_u8,_vg4x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlal_multi_x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_mlal_multi_x4_u16j12svuint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlal_multi_x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_za64,_u16,_vg4x4,,)(slice_base, zn, zm);
+}
+
+// MLSL
+
+// CHECK-LABEL: @test_mlsl_multi_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_mlsl_multi_x4_s8j10svint8x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlsl_multi_x4_s8(uint32_t slice_base, svint8x4_t zn, svint8x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za32,_s8,_vg4x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlsl_multi_x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_mlsl_multi_x4_s16j11svint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlsl_multi_x4_s16(uint32_t slice_base, svint16x4_t zn, svint16x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za64,_s16,_vg4x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlsl_multi_x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_mlsl_multi_x4_u8j11svuint8x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlsl_multi_x4_u8(uint32_t slice_base, svuint8x4_t zn, svuint8x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za32,_u8,_vg4x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_mlsl_multi_x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_mlsl_multi_x4_u16j12svuint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_mlsl_multi_x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_za64,_u16,_vg4x4,,)(slice_base, zn, zm);
+}
+
+// SUMLALL
+
+// CHECK-LABEL: @test_sumlal_multi_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_sumlal_multi_x4_s8j10svint8x4_t11svuint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_sumlal_multi_x4_s8(uint32_t slice_base, svint8x4_t zn, svuint8x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svsumla_za32,_s8,_vg4x4,,)(slice_base, zn, zm);
+}
+
+// USMLALL
+
+// CHECK-LABEL: @test_usmlal_multi_x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_usmlal_multi_x4_u8j11svuint8x4_t10svint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZM]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], <vscale x 16 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_usmlal_multi_x4_u8(uint32_t slice_base, svuint8x4_t zn, svint8x4_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svusmla_za32,_u8,_vg4x4,,)(slice_base, zn, zm);
+}
+
+//
+// Indexed x 1
+//
+
+// SMLAL
+
+// CHECK-LABEL: @test_smlal_lane_x1_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlal_lane_x1_s8ju10__SVInt8_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x1_s8(uint32_t slice_base, svint8_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za32,_s8,_vg4x1,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlal_lane_x1_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.lane.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlal_lane_x1_s16ju11__SVInt16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.lane.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x1_s16(uint32_t slice_base, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za64,_s16,_vg4x1,,)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_smlal_lane_x1_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlal_lane_x1_u8ju11__SVUint8_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x1_u8(uint32_t slice_base, svuint8_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za32,_u8,_vg4x1,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlal_lane_x1_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.lane.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlal_lane_x1_u16ju12__SVUint16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.lane.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x1_u16(uint32_t slice_base, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za64,_u16,_vg4x1,,)(slice_base, zn, zm, 7);
+}
+
+// SMLSL
+
+// CHECK-LABEL: @test_smlsl_lane_x1_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlsl_lane_x1_s8ju10__SVInt8_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x1_s8(uint32_t slice_base, svint8_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za32,_s8,_vg4x1,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlsl_lane_x1_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.lane.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlsl_lane_x1_s16ju11__SVInt16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.lane.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x1_s16(uint32_t slice_base, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za64,_s16,_vg4x1,,)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_smlsl_lane_x1_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlsl_lane_x1_u8ju11__SVUint8_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x1_u8(uint32_t slice_base, svuint8_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za32,_u8,_vg4x1,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlsl_lane_x1_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.lane.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlsl_lane_x1_u16ju12__SVUint16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.lane.vg4x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x1_u16(uint32_t slice_base, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za64,_u16,_vg4x1,,)(slice_base, zn, zm, 7);
+}
+
+// SUMLALL
+
+// CHECK-LABEL: @test_sumlall_lane_x1_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_sumlall_lane_x1_s8ju10__SVInt8_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_sumlall_lane_x1_s8(uint32_t slice_base, svint8_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svsumla_lane_za32,_s8,_vg4x1,,)(slice_base, zn, zm, 15);
+}
+
+// USMLALL
+
+// CHECK-LABEL: @test_usmlall_lane_x1_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_usmlall_lane_x1_u8ju11__SVUint8_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.lane.vg4x1.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_usmlall_lane_x1_u8(uint32_t slice_base, svuint8_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svusmla_lane_za32,_u8,_vg4x1,,)(slice_base, zn, zm, 15);
+}
+
+//
+// Indexed x 2
+//
+
+// SMLAL
+
+// CHECK-LABEL: @test_smlal_lane_x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlal_lane_x2_s8j10svint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x2_s8(uint32_t slice_base, svint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za32,_s8,_vg4x2,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlal_lane_x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.lane.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlal_lane_x2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.lane.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za64,_s16,_vg4x2,,)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_smlal_lane_x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlal_lane_x2_u8j11svuint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x2_u8(uint32_t slice_base, svuint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za32,_u8,_vg4x2,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlal_lane_x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.lane.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlal_lane_x2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.lane.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za64,_u16,_vg4x2,,)(slice_base, zn, zm, 7);
+}
+
+// SMLSL
+
+// CHECK-LABEL: @test_smlsl_lane_x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlsl_lane_x2_s8j10svint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x2_s8(uint32_t slice_base, svint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za32,_s8,_vg4x2,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlsl_lane_x2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.lane.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlsl_lane_x2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.lane.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za64,_s16,_vg4x2,,)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_smlsl_lane_x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlsl_lane_x2_u8j11svuint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x2_u8(uint32_t slice_base, svuint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za32,_u8,_vg4x2,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlsl_lane_x2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.lane.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlsl_lane_x2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.lane.vg4x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za64,_u16,_vg4x2,,)(slice_base, zn, zm, 7);
+}
+
+// SUMLALL
+
+// CHECK-LABEL: @test_sumlall_lane_x2_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_sumlall_lane_x2_s8j10svint8x2_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_sumlall_lane_x2_s8(uint32_t slice_base, svint8x2_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svsumla_lane_za32,_s8,_vg4x2,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_usmlall_lane_x2_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_usmlall_lane_x2_u8j11svuint8x2_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.lane.vg4x2.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_usmlall_lane_x2_u8(uint32_t slice_base, svuint8x2_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svusmla_lane_za32,_u8,_vg4x2,,)(slice_base, zn, zm, 15);
+}
+
+//
+// Indexed x 4
+//
+
+// MLAL
+
+// CHECK-LABEL: @test_smlal_lane_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlal_lane_x4_s8j10svint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x4_s8(uint32_t slice_base, svint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za32,_s8,_vg4x4,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlal_lane_x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.lane.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlal_lane_x4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smla.za64.lane.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za64,_s16,_vg4x4,,)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_smlal_lane_x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlal_lane_x4_u8j11svuint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x4_u8(uint32_t slice_base, svuint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za32,_u8,_vg4x4,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlal_lane_x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.lane.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlal_lane_x4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umla.za64.lane.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlal_lane_x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmla_lane_za64,_u16,_vg4x4,,)(slice_base, zn, zm, 7);
+}
+
+// MLSL
+
+// CHECK-LABEL: @test_smlsl_lane_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlsl_lane_x4_s8j10svint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x4_s8(uint32_t slice_base, svint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za32,_s8,_vg4x4,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlsl_lane_x4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.lane.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlsl_lane_x4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smls.za64.lane.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za64,_s16,_vg4x4,,)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_smlsl_lane_x4_u8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_smlsl_lane_x4_u8j11svuint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x4_u8(uint32_t slice_base, svuint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za32,_u8,_vg4x4,,)(slice_base, zn, zm, 15);
+}
+
+// CHECK-LABEL: @test_smlsl_lane_x4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.lane.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_smlsl_lane_x4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umls.za64.lane.vg4x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_smlsl_lane_x4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svmls_lane_za64,_u16,_vg4x4,,)(slice_base, zn, zm, 7);
+}
+
+// SUMLALL
+
+// CHECK-LABEL: @test_sumlall_lane_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_sumlall_lane_x4_s8j10svint8x4_tu11__SVUint8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.sumla.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_sumlall_lane_x4_s8(uint32_t slice_base, svint8x4_t zn, svuint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svsumla_lane_za32,_s8,_vg4x4,,)(slice_base, zn, zm, 15);
+}
+
+// USMLALL
+
+// CHECK-LABEL: @test_usmlall_lane_x4_s8(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_usmlall_lane_x4_s8j11svuint8x4_tu10__SVInt8_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[ZN]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.usmla.za32.lane.vg4x4.nxv16i8(i32 [[SLICE_BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], <vscale x 16 x i8> [[ZM:%.*]], i32 15)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_usmlall_lane_x4_s8(uint32_t slice_base, svuint8x4_t zn, svint8_t zm) __arm_streaming __arm_shared_za
+{
+  SVE_ACLE_FUNC(svusmla_lane_za32,_u8,_vg4x4,,)(slice_base, zn, zm, 15);
+}
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mls.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mls.c
new file mode 100644
index 000000000000..6830a399e91d
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mls.c
@@ -0,0 +1,292 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sme-f64f64 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED,A5) A1##A3##A5
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3,A4,A5) A1##A2##A3##A4##A5
+#endif
+
+//
+// Multi, multi
+// CHECK-LABEL: @test_svmls2_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZM]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls2_f32j13svfloat32x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZM]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls2_f32(uint32_t slice_base, svfloat32x2_t zn, svfloat32x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_za32,,_f32,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls4_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 4)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], <vscale x 4 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls4_f32j13svfloat32x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZM]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], <vscale x 4 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls4_f32(uint32_t slice_base, svfloat32x4_t zn, svfloat32x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_za32,,_f32,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, single
+// CHECK-LABEL: @test_svmls_single2_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.single.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single2_f32j13svfloat32x2_tu13__SVFloat32_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.single.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single2_f32(uint32_t slice_base, svfloat32x2_t zn, svfloat32_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_single_za32,,_f32,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single4_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.single.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single4_f32j13svfloat32x4_tu13__SVFloat32_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.single.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single4_f32(uint32_t slice_base, svfloat32x4_t zn, svfloat32_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_single_za32,,_f32,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, indexed
+// CHECK-LABEL: @test_svmls_lane2_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.lane.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane2_f32j13svfloat32x2_tu13__SVFloat32_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.lane.vg1x2.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane2_f32(uint32_t slice_base, svfloat32x2_t zn, svfloat32_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_lane_za32,,_f32,,_vg1x2)(slice_base, zn, zm, 3);
+}
+
+// CHECK-LABEL: @test_svmls_lane4_f32(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.lane.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[ZM:%.*]], i32 3)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane4_f32j13svfloat32x4_tu13__SVFloat32_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[ZN]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.lane.vg1x4.nxv4f32(i32 [[SLICE_BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], <vscale x 4 x float> [[ZM:%.*]], i32 3)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane4_f32(uint32_t slice_base, svfloat32x4_t zn, svfloat32_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_lane_za32,,_f32,,_vg1x4)(slice_base, zn, zm, 3);
+}
+
+//
+// Multi, multi
+// CHECK-LABEL: @test_svmls2_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZM]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls2_f64j13svfloat64x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZM]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls2_f64(uint32_t slice_base, svfloat64x2_t zn, svfloat64x2_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_za64,,_f64,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls4_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 2)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], <vscale x 2 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls4_f64j13svfloat64x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZM]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], <vscale x 2 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls4_f64(uint32_t slice_base, svfloat64x4_t zn, svfloat64x4_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_za64,,_f64,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, single
+// CHECK-LABEL: @test_svmls_single2_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.single.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single2_f64j13svfloat64x2_tu13__SVFloat64_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.single.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single2_f64(uint32_t slice_base, svfloat64x2_t zn, svfloat64_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_single_za64,,_f64,,_vg1x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single4_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.single.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single4_f64j13svfloat64x4_tu13__SVFloat64_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.single.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single4_f64(uint32_t slice_base, svfloat64x4_t zn, svfloat64_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_single_za64,,_f64,,_vg1x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, indexed
+// CHECK-LABEL: @test_svmls_lane2_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.lane.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane2_f64j13svfloat64x2_tu13__SVFloat64_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.lane.vg1x2.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane2_f64(uint32_t slice_base, svfloat64x2_t zn, svfloat64_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_lane_za64,,_f64,,_vg1x2)(slice_base, zn, zm, 1);
+}
+
+// CHECK-LABEL: @test_svmls_lane4_f64(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.lane.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[ZM:%.*]], i32 1)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane4_f64j13svfloat64x4_tu13__SVFloat64_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[ZN]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmls.lane.vg1x4.nxv2f64(i32 [[SLICE_BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], <vscale x 2 x double> [[ZM:%.*]], i32 1)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane4_f64(uint32_t slice_base, svfloat64x4_t zn, svfloat64_t zm) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svmls_lane_za64,,_f64,,_vg1x4)(slice_base, zn, zm, 1);
+}
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mlsl.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mlsl.c
new file mode 100644
index 000000000000..0a87d97f649c
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mlsl.c
@@ -0,0 +1,696 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED,A5) A1##A3##A5
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3,A4,A5) A1##A2##A3##A4##A5
+#endif
+
+//
+// Multi, multi
+// CHECK-LABEL: @test_svmls2_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls2_f16j13svfloat16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls2_f16(uint32_t slice_base, svfloat16x2_t zn, svfloat16x2_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_f16,_vg2x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls2_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z16test_svmls2_bf16j14svbfloat16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls2_bf16(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16x2_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_bf16,_vg2x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls2_u16j12svuint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16x2_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_u16,_vg2x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls2_s16j11svint16x2_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls2_s16(uint32_t slice_base, svint16x2_t zn, svint16x2_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_s16,_vg2x2,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls4_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], <vscale x 8 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls4_f16j13svfloat16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], <vscale x 8 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls4_f16(uint32_t slice_base, svfloat16x4_t zn, svfloat16x4_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_f16,_vg2x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls4_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z16test_svmls4_bf16j14svbfloat16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], <vscale x 8 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls4_bf16(uint32_t slice_base, svbfloat16x4_t zn, svbfloat16x4_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_bf16,_vg2x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls4_u16j12svuint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16x4_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_u16,_vg2x4,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z15test_svmls4_s16j11svint16x4_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZM]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], <vscale x 8 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls4_s16(uint32_t slice_base, svint16x4_t zn, svint16x4_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_s16,_vg2x4,,)(slice_base, zn, zm);
+}
+
+//
+// Multi, single
+// CHECK-LABEL: @test_svmls_single1_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x1.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[ZN:%.*]], <vscale x 8 x half> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single1_f16ju13__SVFloat16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x1.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[ZN:%.*]], <vscale x 8 x half> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single1_f16(uint32_t slice_base, svfloat16_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_f16,_vg2x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single1_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x1.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[ZN:%.*]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmls_single1_bf16ju14__SVBfloat16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x1.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[ZN:%.*]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single1_bf16(uint32_t slice_base, svbfloat16_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_bf16,_vg2x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single1_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.single.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single1_u16ju12__SVUint16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.single.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single1_u16(uint32_t slice_base, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_u16,_vg2x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single1_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.single.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single1_s16ju11__SVInt16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.single.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single1_s16(uint32_t slice_base, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_za32,_s16,_vg2x1,,)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single2_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single2_f16j13svfloat16x2_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single2_f16(uint32_t slice_base, svfloat16x2_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_single_za32,,_f16,,_vg2x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single2_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmls_single2_bf16j14svbfloat16x2_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single2_bf16(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_single_za32,,_bf16,,_vg2x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.single.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.single.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_single_za32,,_u16,,_vg2x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.single.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.single.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_single_za32,,_s16,,_vg2x2)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single4_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single4_f16j13svfloat16x4_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single4_f16(uint32_t slice_base, svfloat16x4_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_single_za32,,_f16,,_vg2x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single4_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z23test_svmls_single4_bf16j14svbfloat16x4_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.single.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single4_bf16(uint32_t slice_base, svbfloat16x4_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_single_za32,,_bf16,,_vg2x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.single.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.single.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_single_za32,,_u16,,_vg2x4)(slice_base, zn, zm);
+}
+
+// CHECK-LABEL: @test_svmls_single4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.single.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z22test_svmls_single4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.single.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_single4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_single_za32,,_s16,,_vg2x4)(slice_base, zn, zm);
+}
+
+//
+// Multi, indexed
+//
+
+// CHECK-LABEL: @test_svmls_lane1_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x1.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[ZN:%.*]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane1_f16ju13__SVFloat16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x1.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[ZN:%.*]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane1_f16(uint32_t slice_base, svfloat16_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_f16,,_vg2x1)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane1_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x1.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[ZN:%.*]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_svmls_lane1_bf16ju14__SVBfloat16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x1.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[ZN:%.*]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane1_bf16(uint32_t slice_base, svbfloat16_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_bf16,,_vg2x1)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane1_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.lane.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane1_u16ju12__SVUint16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.lane.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane1_u16(uint32_t slice_base, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_u16,,_vg2x1)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane1_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.lane.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane1_s16ju11__SVInt16_tS_(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.lane.vg2x1.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane1_s16(uint32_t slice_base, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_s16,,_vg2x1)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane2_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane2_f16j13svfloat16x2_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x2.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane2_f16(uint32_t slice_base, svfloat16x2_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_f16,,_vg2x2)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane2_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_svmls_lane2_bf16j14svbfloat16x2_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x2.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane2_bf16(uint32_t slice_base, svbfloat16x2_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_bf16,,_vg2x2)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane2_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.lane.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane2_u16j12svuint16x2_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.lane.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane2_u16(uint32_t slice_base, svuint16x2_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_u16,,_vg2x2)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane2_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.lane.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane2_s16j11svint16x2_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.lane.vg2x2.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane2_s16(uint32_t slice_base, svint16x2_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_s16,,_vg2x2)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane4_f16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane4_f16j13svfloat16x4_tu13__SVFloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x4.nxv8f16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], <vscale x 8 x half> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane4_f16(uint32_t slice_base, svfloat16x4_t zn, svfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_f16,,_vg2x4)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane4_bf16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z21test_svmls_lane4_bf16j14svbfloat16x4_tu14__SVBfloat16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.fmlsl.lane.vg2x4.nxv8bf16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], <vscale x 8 x bfloat> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane4_bf16(uint32_t slice_base, svbfloat16x4_t zn, svbfloat16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_bf16,,_vg2x4)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane4_u16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.lane.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane4_u16j12svuint16x4_tu12__SVUint16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.umlsl.lane.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane4_u16(uint32_t slice_base, svuint16x4_t zn, svuint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_u16,,_vg2x4)(slice_base, zn, zm, 7);
+}
+
+// CHECK-LABEL: @test_svmls_lane4_s16(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.lane.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z20test_svmls_lane4_s16j11svint16x4_tu11__SVInt16_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[ZN]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.smlsl.lane.vg2x4.nxv8i16(i32 [[SLICE_BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], <vscale x 8 x i16> [[ZM:%.*]], i32 7)
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svmls_lane4_s16(uint32_t slice_base, svint16x4_t zn, svint16_t zm) __arm_streaming __arm_shared_za
+{
+   SVE_ACLE_FUNC(svmls_lane_za32,,_s16,,_vg2x4)(slice_base, zn, zm, 7);
+}
diff --git a/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp b/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
index 6ab6dabd92b1..d07d83a53e46 100644
--- a/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
+++ b/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
@@ -241,3 +241,47 @@ void test_bfmlslb_bad_lane(svfloat32_t zda, svbfloat16_t zn, svbfloat16_t zm) __
   svbfmlslb_lane_f32(zda, zn, zm, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
   svbfmlslt_lane_f32(zda, zn, zm, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
 }
+
+void test_multiply_add_sub_long(uint32_t base, svint8_t s8, svuint8_t u8,
+                                svint16_t s16, svuint16_t u16, svint8x2_t s8x2,
+                                svuint8x2_t u8x2, svint16x2_t s16x2, svuint16x2_t u16x2,
+                                svint8x4_t s8x4, svuint8x4_t u8x4, svint16x4_t s16x4, svuint16x4_t u16x4) __arm_streaming __arm_shared_za {
+
+  svmla_lane_za32_s8_vg4x1(base, s8, s8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmla_lane_za32_u8_vg4x1(base, u8, u8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmla_lane_za64_s16_vg4x1(base, s16, s16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svmla_lane_za64_u16_vg4x1(base, u16, u16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+
+  svmla_lane_za32_s8_vg4x2(base, s8x2, s8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmla_lane_za32_u8_vg4x2(base, u8x2, u8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmla_lane_za64_s16_vg4x2(base, s16x2, s16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svmla_lane_za64_u16_vg4x2(base, u16x2, u16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+
+  svmla_lane_za32_s8_vg4x4(base, s8x4, s8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmla_lane_za32_u8_vg4x4(base, u8x4, u8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmla_lane_za64_s16_vg4x4(base, s16x4, s16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svmla_lane_za64_u16_vg4x4(base, u16x4, u16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+
+  svmls_lane_za32_s8_vg4x1(base, s8, s8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmls_lane_za32_u8_vg4x1(base, u8, u8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmls_lane_za64_s16_vg4x1(base, s16, s16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svmls_lane_za64_u16_vg4x1(base, u16, u16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+
+  svmls_lane_za32_s8_vg4x2(base, s8x2, s8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmls_lane_za32_u8_vg4x2(base, u8x2, u8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmls_lane_za64_s16_vg4x2(base, s16x2, s16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svmls_lane_za64_u16_vg4x2(base, u16x2, u16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+
+  svmls_lane_za32_s8_vg4x4(base, s8x4, s8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmls_lane_za32_u8_vg4x4(base, u8x4, u8, 16);   // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svmls_lane_za64_s16_vg4x4(base, s16x4, s16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svmls_lane_za64_u16_vg4x4(base, u16x4, u16, 8); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+
+  svsumla_lane_za32_s8_vg4x1(base, s8, u8, 16);  // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svsumla_lane_za32_s8_vg4x2(base, s8x2, u8, 16); // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svsumla_lane_za32_s8_vg4x4(base, s8x4, u8, 16); // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+
+  svusmla_lane_za32_u8_vg4x1(base, u8, s8, 16);  // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svusmla_lane_za32_u8_vg4x2(base, u8x2, s8, 16); // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+  svusmla_lane_za32_u8_vg4x4(base, u8x4, s8, 16); // expected-error {{argument value 16 is outside the valid range [0, 15]}}
+}