Optimized SourceOver and 16 bit dest fetches, dest stores using NEON.

This makes for example linear gradient blending on top of RGB16
156 % faster (from 20.4 fps to 52.3 fps in my benchmark).

Task-number: QTBUG-6684
Reviewed-by: Gunnar Sletta

1 files changed, 12 insertions, 0 deletions

diff --git a/src/gui/painting/qdrawhelper_neon_p.h b/src/gui/painting/qdrawhelper_neon_p.h
index 6f25243..d6a4509 100644
--- a/src/gui/painting/qdrawhelper_neon_p.h
+++ b/src/gui/painting/qdrawhelper_neon_p.h
@@ -74,6 +74,11 @@ void qt_blend_argb32_on_rgb16_neon(uchar *destPixels, int dbpl,
                                    int w, int h,
                                    int const_alpha);
 
+void qt_blend_argb32_on_argb32_scanline_neon(uint *dest,
+                                             const uint *src,
+                                             int length,
+                                             uint const_alpha);
+
 void qt_blend_rgb16_on_argb32_neon(uchar *destPixels, int dbpl,
                                    const uchar *srcPixels, int sbpl,
                                    int w, int h,
@@ -115,6 +120,13 @@ void qt_transform_image_rgb16_on_rgb16_neon(uchar *destPixels, int dbpl,
                                             const QTransform &targetRectTransform,
                                             int const_alpha);
 
+uint * QT_FASTCALL qt_destFetchRGB16_neon(uint *buffer,
+                                          QRasterBuffer *rasterBuffer,
+                                          int x, int y, int length);
+
+void QT_FASTCALL qt_destStoreRGB16_neon(QRasterBuffer *rasterBuffer,
+                                        int x, int y, const uint *buffer, int length);
+
 #endif // QT_HAVE_NEON
 
 QT_END_NAMESPACE