Re: [FFmpeg-devel] [PATCH 09/11] avcodec/blockdsp: roll-up x86asm preprocessor loop

2017-11-09 Thread Martin Vignali
2017-11-09 12:58 GMT+01:00 James Darnley :

> From: James Darnley 
>
> ---
>  libavcodec/x86/blockdsp.asm | 11 ---
>  1 file changed, 4 insertions(+), 7 deletions(-)
>
> diff --git a/libavcodec/x86/blockdsp.asm b/libavcodec/x86/blockdsp.asm
> index 9d203df8f5..9d0e8a3242 100644
> --- a/libavcodec/x86/blockdsp.asm
> +++ b/libavcodec/x86/blockdsp.asm
> @@ -38,22 +38,19 @@ cglobal clear_block, 1, 1, %1, blocks
>  %assign %%i 0
>  %rep %2
>  mova  [blocksq+mmsize*(0+%%i)], m0
>
>
> do you still need (0+%%i ) ?
Maybe (untested), %%i is enough ?

otherwise lgtm.

Martin
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


[FFmpeg-devel] [PATCH 09/11] avcodec/blockdsp: roll-up x86asm preprocessor loop

2017-11-09 Thread James Darnley
From: James Darnley 

---
 libavcodec/x86/blockdsp.asm | 11 ---
 1 file changed, 4 insertions(+), 7 deletions(-)

diff --git a/libavcodec/x86/blockdsp.asm b/libavcodec/x86/blockdsp.asm
index 9d203df8f5..9d0e8a3242 100644
--- a/libavcodec/x86/blockdsp.asm
+++ b/libavcodec/x86/blockdsp.asm
@@ -38,22 +38,19 @@ cglobal clear_block, 1, 1, %1, blocks
 %assign %%i 0
 %rep %2
 mova  [blocksq+mmsize*(0+%%i)], m0
-mova  [blocksq+mmsize*(1+%%i)], m0
-mova  [blocksq+mmsize*(2+%%i)], m0
-mova  [blocksq+mmsize*(3+%%i)], m0
-%assign %%i %%i+4
+%assign %%i %%i+1
 %endrep
 RET
 %endmacro
 
 INIT_MMX mmx
 %define ZERO pxor
-CLEAR_BLOCK 0, 4
+CLEAR_BLOCK 0, 16
 INIT_XMM sse
 %define ZERO xorps
-CLEAR_BLOCK 1, 2
+CLEAR_BLOCK 1, 8
 INIT_YMM avx
-CLEAR_BLOCK 1, 1
+CLEAR_BLOCK 1, 4
 
 ;-
 ; void ff_clear_blocks(int16_t *blocks);
-- 
2.15.0

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel