[x265] [PATCH] asm: assembly code for pixel_satd_32x24 and rearranged the functions
yuvaraj at multicorewareinc.com
yuvaraj at multicorewareinc.com
Tue Nov 12 07:52:18 CET 2013
# HG changeset patch
# User Yuvaraj Venkatesh <yuvaraj at multicorewareinc.com>
# Date 1384238968 -19800
# Tue Nov 12 12:19:28 2013 +0530
# Node ID 5f77c48448f52d3dfe2d1f7c1dc265df674ff6db
# Parent 1ca01c82609fbb173a665c31adf07c429806d4f1
asm: assembly code for pixel_satd_32x24 and rearranged the functions
diff -r 1ca01c82609f -r 5f77c48448f5 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp Mon Nov 11 15:46:00 2013 +0530
+++ b/source/common/x86/asm-primitives.cpp Tue Nov 12 12:19:28 2013 +0530
@@ -69,7 +69,6 @@
#define HEVC_SATD(cpu) \
HEVC_X64_SATD(cpu) \
p.satd[LUMA_32x32] = cmp<32, 32, 16, 16, x265_pixel_satd_16x16_ ## cpu>; \
- p.satd[LUMA_32x24] = cmp<32, 24, 16, 8, x265_pixel_satd_16x8_ ## cpu>; \
p.satd[LUMA_24x32] = cmp<24, 32, 8, 16, x265_pixel_satd_8x16_ ## cpu>; \
p.satd[LUMA_64x64] = cmp<64, 64, 16, 16, x265_pixel_satd_16x16_ ## cpu>; \
p.satd[LUMA_64x32] = cmp<64, 32, 16, 16, x265_pixel_satd_16x16_ ## cpu>; \
@@ -331,6 +330,7 @@
p.satd[LUMA_12x16] = cmp<12, 16, 4, 16, x265_pixel_satd_4x16_mmx2>;
p.satd[LUMA_32x8] = x265_pixel_satd_32x8_sse2;
p.satd[LUMA_32x16] = x265_pixel_satd_32x16_sse2;
+ p.satd[LUMA_32x24] = x265_pixel_satd_32x24_sse2;
p.sa8d[BLOCK_4x4] = x265_pixel_satd_4x4_mmx2;
p.frame_init_lowres_core = x265_frame_init_lowres_core_mmx2;
@@ -481,6 +481,7 @@
p.satd[LUMA_12x16] = cmp<12, 16, 4, 16, x265_pixel_satd_4x16_sse4>;
p.satd[LUMA_32x8] = x265_pixel_satd_32x8_sse4;
p.satd[LUMA_32x16] = x265_pixel_satd_32x16_sse4;
+ p.satd[LUMA_32x24] = x265_pixel_satd_32x24_sse4;
p.sa8d[BLOCK_8x8] = x265_pixel_sa8d_8x8_sse4;
p.sa8d[BLOCK_16x16] = x265_pixel_sa8d_16x16_sse4;
SA8D_INTER_FROM_BLOCK(sse4);
@@ -499,6 +500,7 @@
p.satd[LUMA_12x16] = cmp<12, 16, 4, 16, x265_pixel_satd_4x16_avx>;
p.satd[LUMA_32x8] = x265_pixel_satd_32x8_avx;
p.satd[LUMA_32x16] = x265_pixel_satd_32x16_avx;
+ p.satd[LUMA_32x24] = x265_pixel_satd_32x24_avx;
p.sa8d[BLOCK_8x8] = x265_pixel_sa8d_8x8_avx;
p.sa8d[BLOCK_16x16] = x265_pixel_sa8d_16x16_avx;
SA8D_INTER_FROM_BLOCK(avx);
diff -r 1ca01c82609f -r 5f77c48448f5 source/common/x86/pixel-a.asm
--- a/source/common/x86/pixel-a.asm Mon Nov 11 15:46:00 2013 +0530
+++ b/source/common/x86/pixel-a.asm Tue Nov 12 12:19:28 2013 +0530
@@ -1780,51 +1780,30 @@
call pixel_satd_16x4_internal
SATD_END_SSE2 m10
-%else
-cglobal pixel_satd_32x16, 4,8,8
-%if WIN64 ;if WIN64 && cpuflag(avx)
- SATD_START_SSE2 m6, m7
+cglobal pixel_satd_32x24, 4,8,8 ;if WIN64 && notcpuflag(avx)
+ SATD_START_SSE2 m10, m7
mov r6, r0
mov r7, r2
- call pixel_satd_8x8_internal
- call pixel_satd_8x8_internal
- lea r0, [r6 + 8]
- lea r2, [r7 + 8]
- call pixel_satd_8x8_internal
- call pixel_satd_8x8_internal
+%if vertical
+ mova m7, [pw_00ff]
+%endif
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
lea r0, [r6 + 16]
lea r2, [r7 + 16]
- call pixel_satd_8x8_internal
- call pixel_satd_8x8_internal
- lea r0, [r6 + 24]
- lea r2, [r7 + 24]
- call pixel_satd_8x8_internal
- call pixel_satd_8x8_internal
- SATD_END_SSE2 m6
-%else ;if !WIN64
- SATD_START_SSE2 m6, m7
- call pixel_satd_8x8_internal
- call pixel_satd_8x8_internal
- mov r0, r0mp
- mov r2, r2mp
- add r0, 8
- add r2, 8
- call pixel_satd_8x8_internal
- call pixel_satd_8x8_internal
- mov r0, r0mp
- mov r2, r2mp
- add r0, 16
- add r2, 16
- call pixel_satd_8x8_internal
- call pixel_satd_8x8_internal
- mov r0, r0mp
- mov r2, r2mp
- add r0, 24
- add r2, 24
- call pixel_satd_8x8_internal
- call pixel_satd_8x8_internal
- SATD_END_SSE2 m6
-%endif
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
+ call pixel_satd_16x4_internal
+ SATD_END_SSE2 m10
+
+%else
cglobal pixel_satd_32x8, 4,6,8 ;if !WIN64
SATD_START_SSE2 m6, m7
@@ -1848,6 +1827,106 @@
call pixel_satd_8x8_internal
SATD_END_SSE2 m6
+%if WIN64
+cglobal pixel_satd_32x16, 4,8,8 ;if WIN64 && cpuflag(avx)
+ SATD_START_SSE2 m6, m7
+ mov r6, r0
+ mov r7, r2
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ lea r0, [r6 + 8]
+ lea r2, [r7 + 8]
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ lea r0, [r6 + 16]
+ lea r2, [r7 + 16]
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ lea r0, [r6 + 24]
+ lea r2, [r7 + 24]
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ SATD_END_SSE2 m6
+%else
+cglobal pixel_satd_32x16, 4,6,8 ;if !WIN64
+ SATD_START_SSE2 m6, m7
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ mov r0, r0mp
+ mov r2, r2mp
+ add r0, 8
+ add r2, 8
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ mov r0, r0mp
+ mov r2, r2mp
+ add r0, 16
+ add r2, 16
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ mov r0, r0mp
+ mov r2, r2mp
+ add r0, 24
+ add r2, 24
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ SATD_END_SSE2 m6
+%endif
+
+%if WIN64
+cglobal pixel_satd_32x24, 4,8,8 ;if WIN64 && cpuflag(avx)
+ SATD_START_SSE2 m6, m7
+ mov r6, r0
+ mov r7, r2
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ lea r0, [r6 + 8]
+ lea r2, [r7 + 8]
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ lea r0, [r6 + 16]
+ lea r2, [r7 + 16]
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ lea r0, [r6 + 24]
+ lea r2, [r7 + 24]
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ SATD_END_SSE2 m6
+%else
+cglobal pixel_satd_32x24, 4,6,8 ;if !WIN64
+ SATD_START_SSE2 m6, m7
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ mov r0, r0mp
+ mov r2, r2mp
+ add r0, 8
+ add r2, 8
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ mov r0, r0mp
+ mov r2, r2mp
+ add r0, 16
+ add r2, 16
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ mov r0, r0mp
+ mov r2, r2mp
+ add r0, 24
+ add r2, 24
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ call pixel_satd_8x8_internal
+ SATD_END_SSE2 m6
+%endif
+
cglobal pixel_satd_16x8, 4,6,8
SATD_START_SSE2 m6, m7
BACKUP_POINTERS
More information about the x265-devel
mailing list