[FFmpeg-devel] [PATCH 09/10] avfilter/vsrc_mandelbrot: use hypot()
gajjanag at mit.edu
Mon Nov 23 19:42:41 CET 2015
On Mon, Nov 23, 2015 at 1:15 PM, Nicolas George <george at nsup.org> wrote:
> Le tridi 3 frimaire, an CCXXIV, Ganesh Ajjanagadde a écrit :
>> I do recall that, but mandelbrot has issues with such benchmarks due
>> to variable number of iterations as I observed while playing with it
>> and as pointed out by Michael. It was hence that I proposed the inner
>> loop benchmark which although it does not get rid of this issue,
>> should reduce the scope of such things.
> I am not sure what this means? Should we understand that two consecutive
> runs of:
> ffmpeg -lavfi mandelbrot -f framecrc -
> ... will not execute in exactly the same way? If so, I think that an option
> for deterministic runs is urgent.
> If not, then I do not see the problem.
Sorry, was vague. The number of iterations depends on the parameters
of the filter. Under fixed settings, one can compare different
benchmarking runs. However, the problem is that changing the
parameters of the filter (and fixing it across runs) may show a
different relative change in performance for the proposed code change
with mandelbrot. The inner loop may remain less affected by such
I guess picking default settings for the filter is reasonable here.
After all, benchmarking is not perfect anyway.
> Nicolas George
More information about the ffmpeg-devel