On 06/05/2011 08:19 PM, Justin Ruggles wrote:

> On 06/05/2011 04:13 PM, Loren Merritt wrote:
> 
>> 6316->5677 cycles on penryn.
> 
> 
> 16258->17385 cycles on athlon64 for the loop changes.  i haven't tested
> the pshufb mask change yet.
> 
> what's the best way to handle such differences?


It's the ALIGN 16 that hurts the SSE2 version on Athlon64 with those
loop changes. I'll remove it and do more testing to see what the effects
are in more cases.

Thanks,
Justin
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to