[x265] [PATCH] asm code for blockcopy_sp, 16xN blocks

[praveen at multicorewareinc.com]

# HG changeset patch
# User Praveen Tiwari
# Date 1383733901 -19800
# Node ID 7486c5c64cc0666deab456aea12a2e1710873680
# Parent  6f9534cb74935fb1593a84a96b495f3ea30ccec4
asm code for blockcopy_sp, 16xN blocks

diff -r 6f9534cb7493 -r 7486c5c64cc0 source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Wed Nov 06 13:06:15 2013 +0530
+++ b/source/common/x86/asm-primitives.cpp	Wed Nov 06 16:01:41 2013 +0530
@@ -339,6 +339,11 @@
         p.chroma_copy_sp[CHROMA_8x6] = x265_blockcopy_sp_8x6_sse2;
         p.chroma_copy_sp[CHROMA_8x8] = x265_blockcopy_sp_8x8_sse2;
         p.chroma_copy_sp[CHROMA_8x16] = x265_blockcopy_sp_8x16_sse2;
+        p.chroma_copy_sp[CHROMA_16x4] = x265_blockcopy_sp_16x4_sse2;
+        p.chroma_copy_sp[CHROMA_16x8] = x265_blockcopy_sp_16x8_sse2;
+        p.chroma_copy_sp[CHROMA_16x12] = x265_blockcopy_sp_16x12_sse2;
+        p.chroma_copy_sp[CHROMA_16x16] = x265_blockcopy_sp_16x16_sse2;
+        p.chroma_copy_sp[CHROMA_16x32] = x265_blockcopy_sp_16x32_sse2;
 #if X86_64
         p.satd[LUMA_8x32] = x265_pixel_satd_8x32_sse2;
         p.satd[LUMA_16x4] = x265_pixel_satd_16x4_sse2;
diff -r 6f9534cb7493 -r 7486c5c64cc0 source/common/x86/blockcopy8.asm
--- a/source/common/x86/blockcopy8.asm	Wed Nov 06 13:06:15 2013 +0530
+++ b/source/common/x86/blockcopy8.asm	Wed Nov 06 16:01:41 2013 +0530
@@ -1146,3 +1146,64 @@
 %endmacro
 
 BLOCKCOPY_SP_W8_H8 8, 16
+
+;-----------------------------------------------------------------------------
+; void blockcopy_sp_%1x%2(pixel *dest, intptr_t destStride, int16_t *src, intptr_t srcStride)
+;-----------------------------------------------------------------------------
+%macro BLOCKCOPY_SP_W16_H4 2
+INIT_XMM sse2
+cglobal blockcopy_sp_%1x%2, 4, 7, 7, dest, destStride, src, srcStride
+
+mov         r6d,    %2
+
+add        r3,      r3
+
+mova       m0,      [tab_Vm]
+
+.loop
+     movu       m1,      [r2]
+     movu       m2,      [r2 + 16]
+     movu       m3,      [r2 + r3]
+     movu       m4,      [r2 + r3 + 16]
+     movu       m5,      [r2 + 2 * r3]
+     movu       m6,      [r2 + 2 * r3 + 16]
+
+     pshufb     m1,      m0
+     pshufb     m2,      m0
+     pshufb     m3,      m0
+     pshufb     m4,      m0
+     pshufb     m5,      m0
+     pshufb     m6,      m0
+
+     movh       [r0],              m1
+     movh       [r0 + 8],          m2
+     movh       [r0 + r1],         m3
+     movh       [r0 + r1 + 8],     m4
+     movh       [r0 + 2 * r1],     m5
+     movh       [r0 + 2 * r1 + 8], m6
+
+     lea        r4,      [r2 + 2 * r3]
+     movu       m1,      [r4 + r3]
+     movu       m2,      [r4 + r3 + 16]
+
+     pshufb     m1,      m0
+     pshufb     m2,      m0
+
+     lea        r5,            [r0 + 2 * r1]
+     movh       [r5 + r1],     m1
+     movh       [r5 + r1 + 8], m2
+
+     lea        r0,              [r5 + 2 * r1]
+     lea        r2,              [r4 + 2 * r3]
+
+     sub        r6d,             4
+     jnz        .loop
+
+RET
+%endmacro
+
+BLOCKCOPY_SP_W16_H4 16,  4
+BLOCKCOPY_SP_W16_H4 16,  8
+BLOCKCOPY_SP_W16_H4 16, 12
+BLOCKCOPY_SP_W16_H4 16, 16
+BLOCKCOPY_SP_W16_H4 16, 32