[FFmpeg-devel] [PATCH v2] avcodec/fft_template: libavcodec/fft_template: improve performance of the ff_fft_init in fft_template

[Paul B Mahol]

On 12/24/18, Michael Niedermayer <michael at niedermayer.cc> wrote:
> On Fri, Dec 21, 2018 at 06:09:50PM +0800, Steven Liu wrote:
>> Before patch:
>> init nbits = 17, get 10000 samples, average cost: 16105 us
>> After patch:
>> init nbits = 17, get 10000 samples, average cost: 15221 us
>>
>> Signed-off-by: Steven Liu <lq at chinaffmpeg.org>
>> ---
>>  libavcodec/fft_template.c | 26 +++++++++++++++-----------
>>  1 file changed, 15 insertions(+), 11 deletions(-)
>>
>> diff --git a/libavcodec/fft_template.c b/libavcodec/fft_template.c
>> index 762c014bc8..2b528be882 100644
>> --- a/libavcodec/fft_template.c
>> +++ b/libavcodec/fft_template.c
>> @@ -261,17 +261,21 @@ av_cold int ff_fft_init(FFTContext *s, int nbits,
>> int inverse)
>>      if (s->fft_permutation == FF_FFT_PERM_AVX) {
>>          fft_perm_avx(s);
>>      } else {
>> -        for(i=0; i<n; i++) {
>> -            int k;
>> -            j = i;
>> -            if (s->fft_permutation == FF_FFT_PERM_SWAP_LSBS)
>> -                j = (j&~3) | ((j>>1)&1) | ((j<<1)&2);
>> -            k = -split_radix_permutation(i, n, s->inverse) & (n-1);
>> -            if (s->revtab)
>> -                s->revtab[k] = j;
>> -            if (s->revtab32)
>> -                s->revtab32[k] = j;
>> -        }
>> +#define SPLIT_RADIX_PERMUTATION(num) do { \
>> +    for(i=0; i<n; i++) {\
>> +        int k;\
>> +        j = i;\
>
>> +        if (s->fft_permutation == FF_FFT_PERM_SWAP_LSBS)\
>
> maybe this if() should be factored out too ?
>
> the change looks good though, and iam not sure this is speed relevant
> enough

Well, it helps. But this code is partially going to be rewritten
anyway in future.