[FFmpeg-devel] [PATCH 2/2] avfilter/avf_showcqt: avoid using fminf()

[James Almer]

On 06/06/14 12:21 PM, Michael Niedermayer wrote:
> On Fri, Jun 06, 2014 at 04:49:21PM +0200, Clément Bœsch wrote:
>> On Fri, Jun 06, 2014 at 04:42:16PM +0200, Michael Niedermayer wrote:
>>> The loop with fminf() changes from 18093856 to 17403218 dezicycles (gcc 4.6.3, sandybridge i7)
>>>
>>> Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
>>> ---
>>>  libavfilter/avf_showcqt.c |    6 +++---
>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/libavfilter/avf_showcqt.c b/libavfilter/avf_showcqt.c
>>> index eaa632d..0870507 100644
>>> --- a/libavfilter/avf_showcqt.c
>>> +++ b/libavfilter/avf_showcqt.c
>>> @@ -357,9 +357,9 @@ static int plot_cqt(AVFilterLink *inlink)
>>>          result[x][2] = r.re * r.re + r.im * r.im;
>>>          result[x][1] = 0.5f * (result[x][0] + result[x][2]);
>>>          result[x][3] = result[x][1];
>>> -        result[x][0] = 255.0f * powf(fminf(1.0f,result[x][0]), g);
>>> -        result[x][1] = 255.0f * powf(fminf(1.0f,result[x][1]), g);
>>> -        result[x][2] = 255.0f * powf(fminf(1.0f,result[x][2]), g);
>>> +        result[x][0] = 255.0f * powf(FFMIN(1.0f,result[x][0]), g);
>>> +        result[x][1] = 255.0f * powf(FFMIN(1.0f,result[x][1]), g);
>>> +        result[x][2] = 255.0f * powf(FFMIN(1.0f,result[x][2]), g);
>>>      }
>>>  
>>
>> LGTM
> 
> patch applied

Either revert the fminf() emulation on libm.h or include lavu/common.h in it, 
because libm.h currently does not have the FFMIN define and it throws errors 
on msvc 2012.