[x265] [PATCH] asm: assembly code for pixel_satd_12x16

[yuvaraj at multicorewareinc.com]

# HG changeset patch
# User Yuvaraj Venkatesh <yuvaraj at multicorewareinc.com>
# Date 1384264561 -19800
#      Tue Nov 12 19:26:01 2013 +0530
# Node ID 794c74c7339734a2fa5c87d05462c76d43adafba
# Parent  76120c6aa908b35734f8e09541090e1834584070
asm: assembly code for pixel_satd_12x16

diff -r 76120c6aa908 -r 794c74c73397 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Tue Nov 12 16:34:37 2013 +0530
+++ b/source/common/x86/asm-primitives.cpp	Tue Nov 12 19:26:01 2013 +0530
@@ -319,7 +319,7 @@
         INIT8(satd, _mmx2);
         HEVC_SATD(mmx2);
         p.satd[LUMA_8x32] = x265_pixel_satd_8x32_sse2;
-        p.satd[LUMA_12x16] = cmp<12, 16, 4, 16, x265_pixel_satd_4x16_mmx2>;
+        p.satd[LUMA_12x16] = x265_pixel_satd_12x16_sse2;
         p.satd[LUMA_16x4] = x265_pixel_satd_16x4_sse2;
         p.satd[LUMA_16x12] = x265_pixel_satd_16x12_sse2;
         p.satd[LUMA_16x32] = x265_pixel_satd_16x32_sse2;
@@ -467,7 +467,7 @@
     if (cpuMask & X265_CPU_SSE4)
     {
         p.satd[LUMA_4x16]   = x265_pixel_satd_4x16_sse4;
-        p.satd[LUMA_12x16]  = cmp<12, 16, 4, 16, x265_pixel_satd_4x16_sse4>;
+        p.satd[LUMA_12x16]  = x265_pixel_satd_12x16_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;
@@ -486,7 +486,7 @@
     {
         p.frame_init_lowres_core = x265_frame_init_lowres_core_avx;
         p.satd[LUMA_4x16]   = x265_pixel_satd_4x16_avx;
-        p.satd[LUMA_12x16]  = cmp<12, 16, 4, 16, x265_pixel_satd_4x16_avx>;
+        p.satd[LUMA_12x16]  = x265_pixel_satd_12x16_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;
diff -r 76120c6aa908 -r 794c74c73397 source/common/x86/pixel-a.asm
--- a/source/common/x86/pixel-a.asm	Tue Nov 12 16:34:37 2013 +0530
+++ b/source/common/x86/pixel-a.asm	Tue Nov 12 19:26:01 2013 +0530
@@ -1993,6 +1993,64 @@
     SATD_END_SSE2 m6, m7
 %endif
 
+%if WIN64
+cglobal pixel_satd_12x16, 4,8,8
+    SATD_START_MMX
+    mov r6, r0
+    mov r7, r2
+%if vertical==0
+    mova m7, [hmul_4p]
+%endif
+    SATD_4x8_SSE vertical, 0, swap
+    lea r0, [r0 + r1*2*SIZEOF_PIXEL]
+    lea r2, [r2 + r3*2*SIZEOF_PIXEL]
+    SATD_4x8_SSE vertical, 1, add
+    lea r0, [r6 + 4]
+    lea r2, [r7 + 4]
+    SATD_4x8_SSE vertical, 1, add
+    lea r0, [r0 + r1*2*SIZEOF_PIXEL]
+    lea r2, [r2 + r3*2*SIZEOF_PIXEL]
+    SATD_4x8_SSE vertical, 1, add
+    lea r0, [r6 + 8]
+    lea r2, [r7 + 8]
+    SATD_4x8_SSE vertical, 1, add
+    lea r0, [r0 + r1*2*SIZEOF_PIXEL]
+    lea r2, [r2 + r3*2*SIZEOF_PIXEL]
+    SATD_4x8_SSE vertical, 1, add
+    HADDW m7, m1
+    movd eax, m7
+    RET
+%else
+cglobal pixel_satd_12x16, 4,6,8
+    SATD_START_MMX
+%if vertical==0
+    mova m7, [hmul_4p]
+%endif
+    SATD_4x8_SSE vertical, 0, swap
+    lea r0, [r0 + r1*2*SIZEOF_PIXEL]
+    lea r2, [r2 + r3*2*SIZEOF_PIXEL]
+    SATD_4x8_SSE vertical, 1, add
+    mov r0, r0mp
+    mov r2, r2mp
+    add r0, 4
+    add r2, 4
+    SATD_4x8_SSE vertical, 1, add
+    lea r0, [r0 + r1*2*SIZEOF_PIXEL]
+    lea r2, [r2 + r3*2*SIZEOF_PIXEL]
+    SATD_4x8_SSE vertical, 1, add
+    mov r0, r0mp
+    mov r2, r2mp
+    add r0, 8
+    add r2, 8
+    SATD_4x8_SSE vertical, 1, add
+    lea r0, [r0 + r1*2*SIZEOF_PIXEL]
+    lea r2, [r2 + r3*2*SIZEOF_PIXEL]
+    SATD_4x8_SSE vertical, 1, add
+    HADDW m7, m1
+    movd eax, m7
+    RET
+%endif
+
 cglobal pixel_satd_8x32, 4,6,8
     SATD_START_SSE2 m6, m7
 %if vertical