Add AVX2 autovectorized versions of premultiply

Following up on using GCC's autovectorizing for faster SSE4.1
premultiply, this patch adds specialized autovectorized versions
of premultiply for AVX2, giving another almost doubling in speed.
To make the speed up for AVX2 and also SSE4_1 available to non-GCC
compilers, the target-specific methods have been moved to separate
files.

Change-Id: I97ce05be67f4adeeb9a096eef80fd5fb662099f3
Reviewed-by: Gunnar Sletta <gunnar@sletta.org>

1 files changed, 16 insertions, 0 deletions

diff --git a/src/gui/painting/qdrawhelper_p.h b/src/gui/painting/qdrawhelper_p.h
index 08bc0776f7..51e51fd53f 100644
--- a/src/gui/painting/qdrawhelper_p.h
+++ b/src/gui/painting/qdrawhelper_p.h
@@ -893,6 +893,22 @@ inline int qBlue565(quint16 rgb) {
     return (b << 3) | (b >> 2);
 }
 
+
+static Q_ALWAYS_INLINE const uint *qt_convertARGB32ToARGB32PM(uint *buffer, const uint *src, int count)
+{
+    for (int i = 0; i < count; ++i)
+        buffer[i] = qPremultiply(src[i]);
+    return buffer;
+}
+
+static Q_ALWAYS_INLINE const uint *qt_convertRGBA8888ToARGB32PM(uint *buffer, const uint *src, int count)
+{
+    for (int i = 0; i < count; ++i)
+        buffer[i] = qPremultiply(RGBA2ARGB(src[i]));
+    return buffer;
+}
+
+
 const uint qt_bayer_matrix[16][16] = {
     { 0x1, 0xc0, 0x30, 0xf0, 0xc, 0xcc, 0x3c, 0xfc,
       0x3, 0xc3, 0x33, 0xf3, 0xf, 0xcf, 0x3f, 0xff},