From 92f441ae863ba00438a8b496a3107a42a2e270ec Mon Sep 17 00:00:00 2001
From: Daniel Kang <daniel.d.kang@gmail.com>
Date: Wed, 29 Dec 2010 21:35:34 +0000
Subject: [PATCH] Port pred4x4_vertical_left_mmxext (H.264 intra prediction)
from x264 (authors: Jason, Loren, Holger) to FFmpeg. Patch by Daniel Kang
<daniel dot d dot kang at gmail com>, as part of Google's GCI 2010.
Originally committed as revision 26155 to svn://svn.ffmpeg.org/ffmpeg/trunk
---
libavcodec/x86/h264_intrapred.asm | 25 +++++++++++++++++++++++++
libavcodec/x86/h264_intrapred_init.c | 3 +++
2 files changed, 28 insertions(+)
diff --git a/libavcodec/x86/h264_intrapred.asm b/libavcodec/x86/h264_intrapred.asm
index b82e02fee1c..81f0fc659f1 100644
--- a/libavcodec/x86/h264_intrapred.asm
+++ b/libavcodec/x86/h264_intrapred.asm
@@ -2536,4 +2536,29 @@ cglobal pred4x4_down_left_mmxext, 3,3
psrlq m0, 8
movd [r1+r2*2], m0
RET
+
+;-----------------------------------------------------------------------------
+; void pred4x4_vertical_left_mmxext(uint8_t *src, const uint8_t *topright, int stride)
+;-----------------------------------------------------------------------------
+
+INIT_MMX
+cglobal pred4x4_vertical_left_mmxext, 3,3
+ sub r0, r2
+ movq m1, [r0]
+ punpckldq m1, [r1]
+ movq m3, m1
+ movq m2, m1
+ psrlq m3, 8
+ psrlq m2, 16
+ movq m4, m3
+ pavgb m4, m1
+ PRED4x4_LOWPASS m0, m1, m2, m3, m5
+ lea r1, [r0+r2*2]
+ movh [r0+r2*1], m4
+ movh [r0+r2*2], m0
+ psrlq m4, 8
+ psrlq m0, 8
+ movh [r1+r2*1], m4
+ movh [r1+r2*2], m0
+ RET
%endif
diff --git a/libavcodec/x86/h264_intrapred_init.c b/libavcodec/x86/h264_intrapred_init.c
index 6367871f2c7..202bbd85932 100644
--- a/libavcodec/x86/h264_intrapred_init.c
+++ b/libavcodec/x86/h264_intrapred_init.c
@@ -84,6 +84,7 @@ void ff_pred8x8l_horizontal_down_sse2(uint8_t *src, int has_topleft, int has_top
void ff_pred8x8l_horizontal_down_ssse3(uint8_t *src, int has_topleft, int has_topright, int stride);
void ff_pred4x4_dc_mmxext (uint8_t *src, const uint8_t *topright, int stride);
void ff_pred4x4_down_left_mmxext (uint8_t *src, const uint8_t *topright, int stride);
+void ff_pred4x4_vertical_left_mmxext(uint8_t *src, const uint8_t *topright, int stride);
void ff_pred4x4_tm_vp8_mmx (uint8_t *src, const uint8_t *topright, int stride);
void ff_pred4x4_tm_vp8_mmxext (uint8_t *src, const uint8_t *topright, int stride);
void ff_pred4x4_tm_vp8_ssse3 (uint8_t *src, const uint8_t *topright, int stride);
@@ -133,6 +134,8 @@ void ff_h264_pred_init_x86(H264PredContext *h, int codec_id)
#if CONFIG_GPL
if (codec_id == CODEC_ID_VP8 || codec_id == CODEC_ID_H264)
h->pred4x4 [DIAG_DOWN_LEFT_PRED ] = ff_pred4x4_down_left_mmxext;
+ if (codec_id == CODEC_ID_SVQ3 || codec_id == CODEC_ID_H264)
+ h->pred4x4 [VERT_LEFT_PRED ] = ff_pred4x4_vertical_left_mmxext;
if (codec_id == CODEC_ID_SVQ3 || codec_id == CODEC_ID_H264) {
h->pred8x8 [TOP_DC_PRED8x8 ] = ff_pred8x8_top_dc_mmxext;
h->pred8x8 [DC_PRED8x8 ] = ff_pred8x8_dc_mmxext;
--
GitLab