[FFmpeg-devel] [PATCH][VAAPI][2/6] Add common data structures and helpers (take 3)

[Gwenole Beauchesne]

On Mon, 9 Mar 2009, Michael Niedermayer wrote:

>> Agner's is indeed the best, then fast_memcpy/mmx (the very old one), then
>> libc, then fast_memcpy/sse2.
>
> could you post the source that was used to generate above numbers?
> At least the mplayer devs might be interrested if agners version is
> faster.

$ cachebench -p (memcpy) + patch attached + add -fno-builtin:
<http://icl.cs.utk.edu/projects/llcbench/>

-Dtest_memcpy=A_memcpy for Agner's
-Dtest_memcpy=fast_memcpy for MPlayer's

Note: for larger blocks (>= 512 KB), fast_memcpy/sse2 was the best on my 
CPU. I have not fully tested unaligned cases though. Agner's should behave 
better since he also has SSSE3 optimizations (PALIGNR). BTW, his lib is 
now GPLv3.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cachebench.test_memcpy.patch
Type: text/x-diff
Size: 592 bytes
Desc: 
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20090309/04fe0fd1/attachment.patch>