From 34ff885bff3703d6e4d8d05192bce6f4d26edb90 Mon Sep 17 00:00:00 2001
From: Allan Sandfeld Jensen <allan.jensen@digia.com>
Date: Wed, 13 Nov 2013 14:07:53 +0100
Subject: [PATCH] Make drawing to and from RGBA8888 images faster

After profiling drawing with RGBA8888 images most of the time appears
to be spend in converting to and from ARGB32PM.

This patch adds four small converters that are 3-4 times faster than the
generic converter.

Change-Id: I3c7498756f440ca3ea9c1417b26dd8e1953b9d06
Reviewed-by: Gunnar Sletta <gunnar.sletta@digia.com>
---
 src/gui/painting/qdrawhelper.cpp | 44 +++++++++++++++++++++++++++-----
 1 file changed, 38 insertions(+), 6 deletions(-)

diff --git a/src/gui/painting/qdrawhelper.cpp b/src/gui/painting/qdrawhelper.cpp
index a037545dc2..8b36a0e28f 100644
--- a/src/gui/painting/qdrawhelper.cpp
+++ b/src/gui/painting/qdrawhelper.cpp
@@ -118,6 +118,22 @@ static const uint *QT_FASTCALL convertARGB32ToARGB32PM(uint *buffer, const uint
     return buffer;
 }
 
+static const uint *QT_FASTCALL convertRGBA8888PMToARGB32PM(uint *buffer, const uint *src, int count,
+                                                           const QPixelLayout *, const QRgb *)
+{
+    for (int i = 0; i < count; ++i)
+        buffer[i] = RGBA2ARGB(src[i]);
+    return buffer;
+}
+
+static const uint *QT_FASTCALL convertRGBA8888ToARGB32PM(uint *buffer, const uint *src, int count,
+                                                         const QPixelLayout *, const QRgb *)
+{
+    for (int i = 0; i < count; ++i)
+        buffer[i] = PREMUL(RGBA2ARGB(src[i]));
+    return buffer;
+}
+
 static const uint *QT_FASTCALL convertToRGB32(uint *buffer, const uint *src, int count,
                                               const QPixelLayout *layout, const QRgb *)
 {
@@ -222,6 +238,22 @@ static const uint *QT_FASTCALL convertARGB32FromARGB32PM(uint *buffer, const uin
     return buffer;
 }
 
+static const uint *QT_FASTCALL convertRGBA8888PMFromARGB32PM(uint *buffer, const uint *src, int count,
+                                                             const QPixelLayout *, const QRgb *)
+{
+    for (int i = 0; i < count; ++i)
+        buffer[i] = ARGB2RGBA(src[i]);
+    return buffer;
+}
+
+static const uint *QT_FASTCALL convertRGBA8888FromARGB32PM(uint *buffer, const uint *src, int count,
+                                                             const QPixelLayout *, const QRgb *)
+{
+    for (int i = 0; i < count; ++i)
+        buffer[i] = ARGB2RGBA(INV_PREMUL(src[i]));
+    return buffer;
+}
+
 static const uint *QT_FASTCALL convertFromARGB32PM(uint *buffer, const uint *src, int count,
                                                    const QPixelLayout *layout, const QRgb *)
 {
@@ -415,13 +447,13 @@ QPixelLayout qPixelLayouts[QImage::NImageFormats] = {
     { 4,  8, 4,  4, 4,  0, 0,  0, false, QPixelLayout::BPP16, convertToRGB32, convertFromARGB32PM }, // Format_RGB444
     { 4,  8, 4,  4, 4,  0, 4, 12,  true, QPixelLayout::BPP16, convertToARGB32PM, convertFromARGB32PM }, // Format_ARGB4444_Premultiplied
 #if Q_BYTE_ORDER == Q_BIG_ENDIAN
-    { 8, 24, 8, 16, 8,  8, 0,  0, false, QPixelLayout::BPP32, convertToRGB32, convertRGBFromARGB32PM }, // Format_RGBX8888
-    { 8, 24, 8, 16, 8,  8, 8,  0, false, QPixelLayout::BPP32, convertToARGB32PM, convertFromARGB32PM }, // Format_RGBA8888
-    { 8, 24, 8, 16, 8,  8, 8,  0,  true, QPixelLayout::BPP32, convertToARGB32PM, convertFromARGB32PM }, // Format_RGBA8888_Premultiplied
+    { 8, 24, 8, 16, 8,  8, 0,  0, false, QPixelLayout::BPP32, convertRGBA8888PMToARGB32PM, convertRGBFromARGB32PM }, // Format_RGBX8888
+    { 8, 24, 8, 16, 8,  8, 8,  0, false, QPixelLayout::BPP32, convertRGBA8888ToARGB32PM, convertRGBA8888FromARGB32PM }, // Format_RGBA8888
+    { 8, 24, 8, 16, 8,  8, 8,  0,  true, QPixelLayout::BPP32, convertRGBA8888PMToARGB32PM, convertRGBA8888PMFromARGB32PM }, // Format_RGBA8888_Premultiplied
 #else
-    { 8,  0, 8,  8, 8, 16, 0, 24, false, QPixelLayout::BPP32, convertToRGB32, convertRGBFromARGB32PM }, // Format_RGBX8888
-    { 8,  0, 8,  8, 8, 16, 8, 24, false, QPixelLayout::BPP32, convertToARGB32PM, convertFromARGB32PM }, // Format_RGBA8888 (ABGR32)
-    { 8,  0, 8,  8, 8, 16, 8, 24,  true, QPixelLayout::BPP32, convertToARGB32PM, convertFromARGB32PM }  // Format_RGBA8888_Premultiplied
+    { 8,  0, 8,  8, 8, 16, 0, 24, false, QPixelLayout::BPP32, convertRGBA8888PMToARGB32PM, convertRGBFromARGB32PM }, // Format_RGBX8888
+    { 8,  0, 8,  8, 8, 16, 8, 24, false, QPixelLayout::BPP32, convertRGBA8888ToARGB32PM, convertRGBA8888FromARGB32PM }, // Format_RGBA8888 (ABGR32)
+    { 8,  0, 8,  8, 8, 16, 8, 24,  true, QPixelLayout::BPP32, convertRGBA8888PMToARGB32PM, convertRGBA8888PMFromARGB32PM }  // Format_RGBA8888_Premultiplied
 #endif
 };