[x265] [PATCH] blockcopy_pp_4x32 asm code: eliminated branch instructions and optimize register uses

[praveen at multicorewareinc.com]

# HG changeset patch
# User Praveen Tiwari
# Date 1423133833 -19800
# Node ID d6c384f3625984a57335c679dbc4fcf37fd4f072
# Parent  7105ee079176c18464982fdc15ed0e0e05f5d015
blockcopy_pp_4x32 asm code: eliminated branch instructions and optimize register uses

diff -r 7105ee079176 -r d6c384f36259 source/common/x86/blockcopy8.asm
--- a/source/common/x86/blockcopy8.asm	Thu Feb 05 16:21:10 2015 +0530
+++ b/source/common/x86/blockcopy8.asm	Thu Feb 05 16:27:13 2015 +0530
@@ -217,47 +217,38 @@
     RET
 
 ;-----------------------------------------------------------------------------
-; void blockcopy_pp_%1x%2(pixel* dst, intptr_t dstStride, const pixel* src, intptr_t srcStride)
+; void blockcopy_pp_4x32(pixel* dst, intptr_t dstStride, const pixel* src, intptr_t srcStride)
 ;-----------------------------------------------------------------------------
-%macro BLOCKCOPY_PP_W4_H8 2
 INIT_XMM sse2
-cglobal blockcopy_pp_%1x%2, 4, 7, 4
-    mov    r4d,    %2/8
-    lea    r5,     [3 * r1]
-    lea    r6,     [3 * r3]
-
-.loop:
+cglobal blockcopy_pp_4x32, 4, 6, 4
+
+    lea     r4,    [3 * r1]
+    lea     r5,    [3 * r3]
+
     movd     m0,     [r2]
     movd     m1,     [r2 + r3]
     movd     m2,     [r2 + 2 * r3]
-    movd     m3,     [r2 + r6]
+    movd     m3,     [r2 + r5]
 
     movd     [r0],          m0
     movd     [r0 + r1],     m1
     movd     [r0 + 2 * r1], m2
-    movd     [r0 + r5],     m3
-
+    movd     [r0 + r4],     m3
+
+%rep 7
     lea      r2,     [r2 + 4 * r3]
     movd     m0,     [r2]
     movd     m1,     [r2 + r3]
     movd     m2,     [r2 + 2 * r3]
-    movd     m3,     [r2 + r6]
+    movd     m3,     [r2 + r5]
 
     lea      r0,            [r0 + 4 * r1]
     movd     [r0],          m0
     movd     [r0 + r1],     m1
     movd     [r0 + 2 * r1], m2
-    movd     [r0 + r5],     m3
-
-    lea       r0,                  [r0 + 4 * r1]
-    lea       r2,                  [r2 + 4 * r3]
-
-    dec       r4d
-    jnz       .loop
+    movd     [r0 + r4],     m3
+%endrep
     RET
-%endmacro
-
-BLOCKCOPY_PP_W4_H8 4, 32
 
 ;-----------------------------------------------------------------------------
 ; void blockcopy_pp_6x8(pixel* dst, intptr_t dstStride, const pixel* src, intptr_t srcStride)