blob: 67eb8553638a8678833c5829aff62875d1b7fd15 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
|
// RUN: %clang_cc1 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -S -o - -emit-llvm %s | FileCheck %s
// Test ARM64 SIMD duplicate lane and n intrinsics
#include <arm_neon.h>
void test_vdup_lane_s64(int64x1_t a1) {
// CHECK-LABEL: test_vdup_lane_s64
vdup_lane_s64(a1, 0);
// CHECK: shufflevector
}
void test_vdup_lane_u64(uint64x1_t a1) {
// CHECK-LABEL: test_vdup_lane_u64
vdup_lane_u64(a1, 0);
// CHECK: shufflevector
}
// uncomment out the following code once scalar_to_vector in the backend
// works (for 64 bit?). Change the "CHECK@" to "CHECK<colon>"
/*
float64x1_t test_vdup_n_f64(float64_t a1) {
// CHECK-LABEL@ test_vdup_n_f64
return vdup_n_f64(a1);
// match that an element is inserted into part 0
// CHECK@ insertelement {{.*, i32 0 *$}}
}
*/
float16x8_t test_vdupq_n_f16(float16_t *a1) {
// CHECK-LABEL: test_vdupq_n_f16
return vdupq_n_f16(*a1);
// match that an element is inserted into parts 0-7. The backend better
// turn that into a single dup instruction
// CHECK: insertelement {{.*, i32 0 *$}}
// CHECK: insertelement {{.*, i32 1 *$}}
// CHECK: insertelement {{.*, i32 2 *$}}
// CHECK: insertelement {{.*, i32 3 *$}}
// CHECK: insertelement {{.*, i32 4 *$}}
// CHECK: insertelement {{.*, i32 5 *$}}
// CHECK: insertelement {{.*, i32 6 *$}}
// CHECK: insertelement {{.*, i32 7 *$}}
}
|
