On 06/02/14 12:30, Janne Grunau wrote:
> From: Christophe Gisquet <[email protected]>
> 
> On 2014-02-06 00:40:50 +0000, Christophe Gisquet wrote:
>> It is currently declared as a macro who is set to inlinable functions,
>> among which a Neon implementation.
>>
>> Add a DSP parameter to the macro, so that the implementation can either
>> be an inline function, or a call to the function found in the context,
>> which is the default.
>>
>> On an Arrandale CPU, gain for an SSE2 function of that inlining vs. a call:
>> - Win32: 29 to 26 cycles
>> - Win64: 25 to 23 cycles
>> ---
>>  libavcodec/arm/dca.h |  5 +++--
>>  libavcodec/dcadec.c  | 16 +++-------------
>>  libavcodec/dcadsp.c  |  9 +++++++++
>>  libavcodec/dcadsp.h  |  4 ++++
>>  4 files changed, 19 insertions(+), 15 deletions(-)
> 
> I though more of something like below which is smaller and has imho
> clearer preprocessor use.
> 
> Janne
> 

av_unused might be added.

That way it should not make a difference between Christophe macro and
this speedwise.

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

Reply via email to