Add AVX2 versions of qt_memfill32 and qt_memfill64

The implementation is almost the same 4-way-unrolled loop, but because
of the wider registers, we fill 128 bytes per loop. Unlike the SSE2
implementation, the AVX2 version uses unaligned stores and won't try to
align in the prologue, matching glibc's __memset_avx2 (also unaligned).

Change-Id: Iba4b5c183776497d8ee1fffd15637ccb2a7b83bc
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>

1 files changed, 52 insertions, 1 deletions

diff --git a/src/gui/painting/qdrawhelper_avx2.cpp b/src/gui/painting/qdrawhelper_avx2.cpp
index ec6643deed..adf11999bc 100644
--- a/src/gui/painting/qdrawhelper_avx2.cpp
+++ b/src/gui/painting/qdrawhelper_avx2.cpp
@@ -1,6 +1,7 @@
 /****************************************************************************
 **
-** Copyright (C) 2016 The Qt Company Ltd.
+** Copyright (C) 2018 The Qt Company Ltd.
+** Copyright (C) 2018 Intel Corporation.
 ** Contact: https://www.qt.io/licensing/
 **
 ** This file is part of the QtGui module of the Qt Toolkit.
@@ -38,6 +39,7 @@
 ****************************************************************************/
 
 #include "qdrawhelper_p.h"
+#include "qdrawhelper_x86_p.h"
 #include "qdrawingprimitive_sse2_p.h"
 #include "qrgba64_p.h"
 
@@ -316,6 +318,55 @@ void qt_blend_rgb32_on_rgb32_avx2(uchar *destPixels, int dbpl,
     }
 }
 
+static Q_NEVER_INLINE
+void Q_DECL_VECTORCALL qt_memfillXX_avx2(uchar *dest, __m256i value256, qsizetype bytes)
+{
+    __m128i value128 = _mm256_castsi256_si128(value256);
+
+    // main body
+    __m256i *dst256 = reinterpret_cast<__m256i *>(dest);
+    uchar *end = dest + bytes;
+    while (reinterpret_cast<uchar *>(dst256 + 4) <= end) {
+        _mm256_storeu_si256(dst256 + 0, value256);
+        _mm256_storeu_si256(dst256 + 1, value256);
+        _mm256_storeu_si256(dst256 + 2, value256);
+        _mm256_storeu_si256(dst256 + 3, value256);
+        dst256 += 4;
+    }
+
+    // first epilogue: fewer than 128 bytes / 32 entries
+    bytes = end - reinterpret_cast<uchar *>(dst256);
+    switch (bytes / sizeof(value256)) {
+    case 3: _mm256_storeu_si256(dst256++, value256); Q_FALLTHROUGH();
+    case 2: _mm256_storeu_si256(dst256++, value256); Q_FALLTHROUGH();
+    case 1: _mm256_storeu_si256(dst256++, value256);
+    }
+
+    // second epilogue: fewer than 32 bytes
+    __m128i *dst128 = reinterpret_cast<__m128i *>(dst256);
+    if (bytes & sizeof(value128))
+        _mm_storeu_si128(dst128++, value128);
+
+    // third epilogue: fewer than 16 bytes
+    if (bytes & 8)
+        _mm_storel_epi64(reinterpret_cast<__m128i *>(end - 8), value128);
+}
+
+void qt_memfill64_avx2(quint64 *dest, quint64 value, qsizetype count)
+{
+    qt_memfillXX_avx2(reinterpret_cast<uchar *>(dest), _mm256_set1_epi64x(value), count * sizeof(quint64));
+}
+
+void qt_memfill32_avx2(quint32 *dest, quint32 value, qsizetype count)
+{
+    if (count % 2) {
+        // odd number of pixels, round to even
+        *dest++ = value;
+        --count;
+    }
+    qt_memfillXX_avx2(reinterpret_cast<uchar *>(dest), _mm256_set1_epi32(value), count * sizeof(quint32));
+}
+
 void QT_FASTCALL comp_func_SourceOver_avx2(uint *destPixels, const uint *srcPixels, int length, uint const_alpha)
 {
     Q_ASSERT(const_alpha < 256);