Hi Lance, On Sat, May 6, 2023 at 7:32 AM Lance Wang <lance.lmw...@gmail.com> wrote: > FYI, on my development system, I run two time for the original and modified > version and no obvious difference:
Simply running "time" against the binary isn't an accurate way to measure a 60ms difference for a single frame being processed. For any such execution of ffmpeg the bulk of the time is spent loading the application and in this case loading the 20MB file from disk into memory. In my case I added instrumentation to the decoder to measure how much time it took to perform the actual decode operation. I discarded the patch already since it was like six lines of code and I did the work several weeks ago, but if there is really a dispute about the performance benefit I can obviously recreate it (as can anyone who wants to benchmark it themselves). You would just have to insert a couple of calls to gettimeofday() in libavcodec/bitpacked_dec.c before and after the decoding operation. This is one of those cases where you won't notice the performance difference doing any operation once, but it becomes important when you're processing a live RTP stream of 2160p59 video at 10.4 Gbps. Devin -- Devin Heitmueller, Senior Software Engineer LTN Global Communications o: +1 (301) 363-1001 w: https://ltnglobal.com e: devin.heitmuel...@ltnglobal.com _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
