[x265] [PATCH] blockcopy_pp_8x6: optimize register uses
praveen at multicorewareinc.com
praveen at multicorewareinc.com
Tue Feb 3 12:45:04 CET 2015
# HG changeset patch
# User Praveen Tiwari
# Date 1422963895 -19800
# Node ID bfc9a2d99e20568cb43d9fba0133735009793b00
# Parent d212ce9fa3705b9e7d4d23f14412bd28fe3bbfde
blockcopy_pp_8x6: optimize register uses
diff -r d212ce9fa370 -r bfc9a2d99e20 source/common/x86/blockcopy8.asm
--- a/source/common/x86/blockcopy8.asm Tue Feb 03 14:30:42 2015 +0530
+++ b/source/common/x86/blockcopy8.asm Tue Feb 03 17:14:55 2015 +0530
@@ -327,24 +327,24 @@
; void blockcopy_pp_8x6(pixel* dst, intptr_t dstStride, const pixel* src, intptr_t srcStride)
;-----------------------------------------------------------------------------
INIT_XMM sse2
-cglobal blockcopy_pp_8x6, 4, 7, 6
+cglobal blockcopy_pp_8x6, 4, 4, 6
movh m0, [r2]
movh m1, [r2 + r3]
- movh m2, [r2 + 2 * r3]
- lea r5, [r2 + 2 * r3]
- movh m3, [r5 + r3]
- movh m4, [r5 + 2 * r3]
- lea r5, [r5 + 2 * r3]
- movh m5, [r5 + r3]
-
- movh [r0], m0
- movh [r0 + r1], m1
- movh [r0 + 2 * r1], m2
- lea r6, [r0 + 2 * r1]
- movh [r6 + r1], m3
- movh [r6 + 2 * r1], m4
- lea r6, [r6 + 2 * r1]
- movh [r6 + r1], m5
+ lea r2, [r2 + 2 * r3]
+ movh m2, [r2]
+ movh m3, [r2 + r3]
+ lea r2, [r2 + 2 * r3]
+ movh m4, [r2]
+ movh m5, [r2 + r3]
+
+ movh [r0], m0
+ movh [r0 + r1], m1
+ lea r0, [r0 + 2 * r1]
+ movh [r0], m2
+ movh [r0 + r1], m3
+ lea r0, [r0 + 2 * r1]
+ movh [r0], m4
+ movh [r0 + r1], m5
RET
;-----------------------------------------------------------------------------
More information about the x265-devel
mailing list