Hello,
I have:
driver 460.32.03
cuda 11.2
Quadro RTX 4000
When I set ulMaxDisplayDelay=0, I can see a latency drop for h264 decoding but when I set it to any non-zero value (tried, 1, 2, 4) I get an immediate latency jump to an exactly same value (20ms->80ms for 60fps). It seems that it doesn’t matter what non-zero value I set there. Is it expected?
It’s true that I am trying this through ffmpeg’s api by recompiling it with the desired value: FFmpeg/cuviddec.c at master · FFmpeg/FFmpeg · GitHub but I don’t think that the ffmpeg is causing this.
Is it possible that CUVIDPARSERPARAMS::ulMaxDisplayDelay is, in fact, interpreted as a boolean?