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

2018-12-17 Thread Michael Niedermayer
On Mon, Dec 17, 2018 at 05:13:05PM +0800, Steven Liu wrote:
> After patch:
> init nbits = 17, get 1 samples, duration: 15221
> Before patch:
> init nbits = 17, get 1 samples, duration: 16105
> 
> test script:
> DURATION=0
> for((i=0;i<1;i++)) do
> ./libavcodec/tests/fft -n 17 &>output
> T_DURATION=`grep "duration" output | awk -F"=" '{ print $2}'`
> DURATION=`expr $DURATION + $T_DURATION`
> done
> TOTAL=`expr $DURATION / 1`
> echo $TOTAL
> 
> Signed-off-by: Steven Liu 
> ---
>  libavcodec/fft_template.c | 27 +--
>  1 file changed, 17 insertions(+), 10 deletions(-)

this breaks fate
make: *** [fate-unknown_layout-ac3] Error 1
make: *** [fate-acodec-dca] Error 1
make: *** [fate-acodec-dca2] Error 1
make: *** [fate-lavf-rm] Error 1
make: *** [fate-ac3-fixed-2.0] Error 1
make: *** [fate-ac3-fixed-4.0-downmix-mono] Error 1
make: *** [fate-ac3-fixed-5.1-downmix-stereo] Error 1
make: *** [fate-ac3-fixed-5.1-downmix-mono] Error 1
make: *** [fate-ac3-fixed-encode] Error 1
make: *** [fate-atrac1-1] Error 1
make: *** [fate-atrac3p-2] Error 1
make: *** [fate-atrac3p-1] Error 1
make: *** [fate-opus-testvector01] Error 1
make: *** [fate-opus-testvector06] Error 1
make: *** [fate-opus-testvector05] Error 1
make: *** [fate-opus-testvector11] Error 1
make: *** [fate-opus-testvector08] Error 1
make: *** [fate-opus-tron.6ch.tinypkts] Error 1
make: *** [fate-opus-testvector07] Error 1
make: *** [fate-opus-testvector09] Error 1
make: *** [fate-opus-testvector10] Error 1
make: *** [fate-prores-gray] Error 1
make: *** [fate-aac-al04sf_48] Error 1
make: *** [fate-aac-fixed-al05_44] Error 1
make: *** [fate-aac-er_eld2100np_48_ep0] Error 1
make: *** [fate-aac-fixed-al06_44] Error 1
make: *** [fate-aac-fixed-al04_44] Error 1
make: *** [fate-aac-fixed-al15_44] Error 1
make: *** [fate-aac-fixed-al17_44] Error 1
make: *** [fate-aac-fixed-er_ad6000np_44_ep0] Error 1
make: *** [fate-aac-fixed-al18_44] Error 1
make: *** [fate-aac-fixed-ap05_48] Error 1
make: *** [fate-aac-fixed-al_sbr_hq_cm_48_2] Error 1
make: *** [fate-aac-fixed-er_eld1001np_44_ep0] Error 1
make: *** [fate-aac-fixed-er_eld2000np_48_ep0] Error 1
make: *** [fate-aac-fixed-al_sbr_hq_sr_48_2_fsaac48] Error 1
make: *** [fate-aac-fixed-al_sbr_hq_cm_48_5.1] Error 1
make: *** [fate-dca-core_51_24_48_768_1] Error 1
make: *** [fate-dca-core_51_24_48_768_0] Error 1
make: *** [fate-dca-core_51_24_48_768_1-dmix_2] Error 1
make: *** [fate-dca-xxch_71_24_48_2046] Error 1
make: *** [fate-dca-xbr_xxch_71_24_48_3840] Error 1
make: *** [fate-dca-xch_61_24_48_768-dmix_6] Error 1
make: *** [fate-dca-xbr_xch_61_24_48_3840] Error 1
make: *** [fate-dca-xch_61_24_48_768] Error 1
make: *** [fate-dca-xbr_51_24_48_3840] Error 1
make: *** [fate-dts_es] Error 1
make: *** [fate-dca-core] Error 1
...

[...]
-- 
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

If you fake or manipulate statistics in a paper in physics you will never
get a job again.
If you fake or manipulate statistics in a paper in medicin you will get
a job for life at the pharma industry.


signature.asc
Description: PGP signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


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

2018-12-17 Thread Carl Eugen Hoyos
2018-12-17 15:56 GMT+01:00, Steven Liu :
>
>
>> On Dec 17, 2018, at 22:05, Carl Eugen Hoyos  wrote:
>>
>> 2018-12-17 10:13 GMT+01:00, Steven Liu :
>>> After patch:
>>> init nbits = 17, get 1 samples, duration: 15221
>>> Before patch:
>>> init nbits = 17, get 1 samples, duration: 16105
>>>
>>> test script:
>>> DURATION=0
>>> for((i=0;i<1;i++)) do
>>> ./libavcodec/tests/fft -n 17 &>output
>>> T_DURATION=`grep "duration" output | awk -F"=" '{ print $2}'`
>>> DURATION=`expr $DURATION + $T_DURATION`
>>> done
>>> TOTAL=`expr $DURATION / 1`
>>> echo $TOTAL
>>
>> This script does not allow to reproduce afaict.
>> (There is no "duration" here in output)
>
> Do you mean i should commit 2/3 patch of the libavcodec/tests/fft.c modify?
> or just upload the modify diff here?

I believe your commit message should not contain a script
that cannot be used to reproduce your findings.

Carl Eugen
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


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

2018-12-17 Thread Steven Liu


> On Dec 17, 2018, at 22:05, Carl Eugen Hoyos  wrote:
> 
> 2018-12-17 10:13 GMT+01:00, Steven Liu :
>> After patch:
>> init nbits = 17, get 1 samples, duration: 15221
>> Before patch:
>> init nbits = 17, get 1 samples, duration: 16105
>> 
>> test script:
>> DURATION=0
>> for((i=0;i<1;i++)) do
>> ./libavcodec/tests/fft -n 17 &>output
>> T_DURATION=`grep "duration" output | awk -F"=" '{ print $2}'`
>> DURATION=`expr $DURATION + $T_DURATION`
>> done
>> TOTAL=`expr $DURATION / 1`
>> echo $TOTAL
> 
> This script does not allow to reproduce afaict.
> (There is no "duration" here in output)
Do you mean i should commit 2/3 patch of the libavcodec/tests/fft.c modify?
or just upload the modify diff here?
> 
> Carl Eugen
> ___
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Thanks
Steven





___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


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

2018-12-17 Thread Carl Eugen Hoyos
2018-12-17 10:13 GMT+01:00, Steven Liu :
> After patch:
> init nbits = 17, get 1 samples, duration: 15221
> Before patch:
> init nbits = 17, get 1 samples, duration: 16105
>
> test script:
> DURATION=0
> for((i=0;i<1;i++)) do
> ./libavcodec/tests/fft -n 17 &>output
> T_DURATION=`grep "duration" output | awk -F"=" '{ print $2}'`
> DURATION=`expr $DURATION + $T_DURATION`
> done
> TOTAL=`expr $DURATION / 1`
> echo $TOTAL

This script does not allow to reproduce afaict.
(There is no "duration" here in output)

Carl Eugen
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel