[x265] [PATCH 2 of 2] asm: fix sse_pp[32x64] sse2 asm for 12 bit

[chen]

At 2015-09-03 16:22:53,ramya at multicorewareinc.com wrote:
># HG changeset patch
># User Ramya Sriraman <ramya at multicorewareinc.com>
># Date 1441088473 -19800
>#      Tue Sep 01 11:51:13 2015 +0530
># Node ID 8864f6a4d9ce4c1aa3c1b3c934befffb616ace35
># Parent  83dc8aea6ba7c10e0d78ec7dc34b3d8f7d114563
>asm: fix sse_pp[32x64] sse2 asm for 12 bit
>
>diff -r 83dc8aea6ba7 -r 8864f6a4d9ce source/common/x86/asm-primitives.cpp
>--- a/source/common/x86/asm-primitives.cpp	Wed Aug 26 17:06:25 2015 +0530
>+++ b/source/common/x86/asm-primitives.cpp	Tue Sep 01 11:51:13 2015 +0530
>@@ -1001,11 +1001,11 @@
>         p.chroma[X265_CSP_I422].cu[BLOCK_422_4x8].sse_pp = (pixel_sse_t)PFX(pixel_ssd_ss_4x8_mmx2);
>         p.chroma[X265_CSP_I422].cu[BLOCK_422_8x16].sse_pp = (pixel_sse_t)PFX(pixel_ssd_ss_8x16_sse2);
>         p.chroma[X265_CSP_I422].cu[BLOCK_422_16x32].sse_pp = (pixel_sse_t)PFX(pixel_ssd_ss_16x32_sse2);
>-
>+        p.chroma[X265_CSP_I422].cu[BLOCK_422_32x64].sse_pp = (pixel_sse_t)PFX(pixel_ssd_ss_32x64_sse2);
>+ 
> #if X265_DEPTH <= 10
>         p.cu[BLOCK_4x4].sse_ss = PFX(pixel_ssd_ss_4x4_mmx2);
>         ALL_LUMA_CU(sse_ss, pixel_ssd_ss, sse2);
>-        p.chroma[X265_CSP_I422].cu[BLOCK_422_32x64].sse_pp = (pixel_sse_t)PFX(pixel_ssd_ss_32x64_sse2);
> #endif
> 
>         p.cu[BLOCK_4x4].dct = PFX(dct4_sse2);
>diff -r 83dc8aea6ba7 -r 8864f6a4d9ce source/common/x86/ssd-a.asm
>--- a/source/common/x86/ssd-a.asm	Wed Aug 26 17:06:25 2015 +0530
>+++ b/source/common/x86/ssd-a.asm	Tue Sep 01 11:51:13 2015 +0530
>@@ -125,6 +125,61 @@
>     RET
> %endmacro
> 
>+; Function to find ssd for 32x16 block, sse2, 12 bit depth
>+; Defined sepeartely to be called from SSD_ONE_32 macro
>+INIT_XMM sse2
>+cglobal ssd_ss_32x16 
>+    pxor    m8, m8
>+    mov     r4d, 16
>+.loop:
>+    movu        m0, [r0]
>+    movu        m1, [r0+mmsize]
>+    movu        m2, [r0+2*mmsize]
>+    movu        m3, [r0+3*mmsize]
>+    movu        m4, [r2]
>+    movu        m5, [r2+mmsize]
>+    movu        m6, [r2+2*mmsize]
>+    movu        m7, [r2+3*mmsize]
>+    psubw       m0, m4
>+    psubw       m1, m5 
>+    psubw       m2, m6
>+    psubw       m3, m7
>+    lea         r0, [r0+r1]
>+    lea         r2, [r2+r3]
>+    pmaddwd     m0, m0
>+    pmaddwd     m1, m1
>+    pmaddwd     m2, m2
>+    pmaddwd     m3, m3
>+    paddd       m2, m3
>+    paddd       m1, m2
>+    paddd       m0, m1
>+    paddd       m8, m0
2 \ 
3 + - 0 - 1 - 8
Above dependency link a little longer, try below one
0 \
1 + - + 8
2 + /
3 /

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20150904/677ab3ea/attachment.html>