[x265] [PATCH] blockfill_s_32x32 sse2 asm code optimization

[praveen at multicorewareinc.com]

# HG changeset patch
# User Praveen Tiwari
# Date 1422878196 -19800
# Branch stable
# Node ID c0d8c77ef0da266a08d1eba2d80d05d0ad1d28fb
# Parent  8d03acd70332ccf642fc7222bf6f9e7f005983ba
blockfill_s_32x32 sse2 asm code optimization

optimized LEA instruction

diff -r 8d03acd70332 -r c0d8c77ef0da source/common/x86/blockcopy8.asm
--- a/source/common/x86/blockcopy8.asm	Mon Feb 02 17:13:34 2015 +0530
+++ b/source/common/x86/blockcopy8.asm	Mon Feb 02 17:26:36 2015 +0530
@@ -1855,13 +1855,14 @@
 ;-----------------------------------------------------------------------------
 ; void blockfill_s_%1x%2(int16_t* dst, intptr_t dstride, int16_t val)
 ;-----------------------------------------------------------------------------
-%macro BLOCKFILL_S_W32_H4 2
+%macro BLOCKFILL_S_W32_H8 2
 INIT_XMM sse2
 cglobal blockfill_s_%1x%2, 3, 5, 1, dst, dstStride, val
 
-mov        r3d,           %2/4
+mov        r3d,           %2/8
 
 add        r1,            r1
+lea        r4,            [3 * r1]
 
 movd       m0,            r2d
 pshuflw    m0,            m0,       0
@@ -1883,12 +1884,31 @@
      movu       [r0 + 2 * r1 + 32], m0
      movu       [r0 + 2 * r1 + 48], m0
 
-     lea        r4,                 [r0 + 2 * r1]
-
-     movu       [r4 + r1],          m0
-     movu       [r4 + r1 + 16],     m0
-     movu       [r4 + r1 + 32],     m0
-     movu       [r4 + r1 + 48],     m0
+     movu       [r0 + r4],          m0
+     movu       [r0 + r4 + 16],     m0
+     movu       [r0 + r4 + 32],     m0
+     movu       [r0 + r4 + 48],     m0
+
+     lea        r0,                 [r0 + 4 * r1]
+     movu       [r0],               m0
+     movu       [r0 + 16],          m0
+     movu       [r0 + 32],          m0
+     movu       [r0 + 48],          m0
+
+     movu       [r0 + r1],          m0
+     movu       [r0 + r1 + 16],     m0
+     movu       [r0 + r1 + 32],     m0
+     movu       [r0 + r1 + 48],     m0
+
+     movu       [r0 + 2 * r1],      m0
+     movu       [r0 + 2 * r1 + 16], m0
+     movu       [r0 + 2 * r1 + 32], m0
+     movu       [r0 + 2 * r1 + 48], m0
+
+     movu       [r0 + r4],          m0
+     movu       [r0 + r4 + 16],     m0
+     movu       [r0 + r4 + 32],     m0
+     movu       [r0 + r4 + 48],     m0
 
      lea        r0,                 [r0 + 4 * r1]
 
@@ -1898,7 +1918,7 @@
 RET
 %endmacro
 
-BLOCKFILL_S_W32_H4 32, 32
+BLOCKFILL_S_W32_H8 32, 32
 
 INIT_YMM avx2
 cglobal blockfill_s_32x32, 3, 4, 1