Hi,

On Thu, Apr 12, 2012 at 8:34 PM, Justin Ruggles
<justin.rugg...@gmail.com> wrote:
> On 04/12/2012 04:49 PM, Ronald S. Bultje wrote:
>> On Thu, Apr 12, 2012 at 1:22 PM, Justin Ruggles
>> <justin.rugg...@gmail.com> wrote:
>>> Fixes ac3-encode and eac3-encode FATE test failures with SSE2 disabled.
>>> ---
>>>  libavcodec/x86/ac3dsp.asm |    6 ++++--
>>>  1 files changed, 4 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/libavcodec/x86/ac3dsp.asm b/libavcodec/x86/ac3dsp.asm
>>> index 746fd83..1438811 100644
>>> --- a/libavcodec/x86/ac3dsp.asm
>>> +++ b/libavcodec/x86/ac3dsp.asm
>>> @@ -223,7 +223,8 @@ cglobal float_to_fixed24_3dnow, 3,3,0, dst, src, len
>>>     add  dstq, 32
>>>     sub  lend, 8
>>>     ja .loop
>>> -    REP_RET
>>> +    femms
>>> +    RET
>>>
>>>  INIT_XMM
>>>  cglobal float_to_fixed24_sse, 3,3,3, dst, src, len
>>> @@ -247,7 +248,8 @@ cglobal float_to_fixed24_sse, 3,3,3, dst, src, len
>>>     add      dstq, 32
>>>     sub      lend, 8
>>>     ja .loop
>>> -    REP_RET
>>> +    emms
>>> +    RET
>>
>> OK.
>>
>> (How much does that slow it down compared to the C function?)
>
>
> on Athlon64, best of 5 runs:
>
> C - 22716 cycles
>
> without patch:
> SSE   - 8074
> 3DNow - 7237
>
> with patch:
> SSE   - 8089
> 3DNow - 7249

OK, pushed.

Ronald
_______________________________________________
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to