[Ffmpeg-devel-irc] ffmpeg-devel.log.20190618

[burek]

[00:56:50 CEST] <cone-655> ffmpeg 03Jun Zhao 07release/4.1:72f03b2af489: lavc/libaomenc: Add a maximum constraint of 64 encoder threads.
[04:54:56 CEST] <cone-866> ffmpeg 03Limin Wang 07master:268ab17c519f: libavcodec/videotoolboxenc: Fix compilation broken on macOS 10.12
[12:31:09 CEST] <faeez> Hi, I am planning on adding support to set different PID's to audio and video streams via AVOptions. Direction on how to  would be much appreciated
[12:31:22 CEST] <faeez> In mpegTS
[12:36:15 CEST] <funman> faeez: you can already set AVStream->id no ?
[12:38:05 CEST] <faeez> Not sure of that, as I want to expose the option to the user without making any changes in fftools, I am planning on just making changes in mpegtsenc.c
[12:38:53 CEST] <faeez> also I want to expose few other cutomized options
[12:43:14 CEST] <funman> ffmpeg -i foo.ts -vc copy -acodec copy -streamid 0:42 -streamid 1:444 test.ts
[12:51:44 CEST] <faeez> Thanks funman
[12:55:43 CEST] <faeez> How about handling multiple program, setting different PCR PID, PMTPID, managing time between PCRs for each program
[13:04:32 CEST] <funman> there is an option for pmt pid already
[13:07:41 CEST] <faeez> I am assuming you are suggesting 'mpegts_pmt_start_pid'. My intention is to provide exclusive control to user to set different pmt PID's for each program. 
[13:08:16 CEST] <faeez> mpegts_pmt_start_pid only allows to set the first pid of pmt
[13:10:13 CEST] <funman> yes that's the one i saw
[13:12:03 CEST] <faeez> Are you aware of anyway to have dynamic array for setting avoptions?
[13:13:11 CEST] <funman> no
[13:13:12 CEST] <faeez> other then OptionDef exposed by ffmpeg_opt.c
[13:14:55 CEST] <faeez> Appreciate your time
[15:24:00 CEST] <cone-045> ffmpeg 03Gyan Doshi 07master:2fdbeb0b8cc3: avformat/segment: fix increment_tc
[15:59:22 CEST] <mkver> jkqxz: I am currently rebasing the old patchset to update the AVCodecParameters with the *_metadata filters. Do you want them to be updated optionally (but by default) or should it be done totally automatically (as James suggested)?
[17:29:00 CEST] <cone-045> ffmpeg 03Jun Zhao 07master:ebcf4d354f5c: lavfi/af_asetnsamples: Remove the redundant condition check
[22:41:09 CEST] <Marshalleq> Hello, I have read that ffmpeg becomes less efficient as you add more cores.  Is this still the case in 2019?  If so how many cores per ffmpeg instance would be optimum?
[22:49:08 CEST] <durandal_1707> Marshalleq: where have you read that lie?
[22:50:11 CEST] <Marshalleq> durandal_1707 I'll see if I can ferret it out - I couldn't find anything that said otherwise, so I thought I'd ask here.
[22:50:17 CEST] <JEEB> well we do say that if you set AVCodecContext's threads to over 16 that it is not recommend etc etc
[22:50:29 CEST] <JEEB> although you can still set it much much higher
[22:50:39 CEST] <JEEB> granted, it might not be efficient in many ways :P
[22:51:16 CEST] <JEEB> but saying taht FFmpeg in general would become less efficient... less true. since in many cases you are running more than one thing
[22:51:35 CEST] <JEEB> so you can run your decoder with X threads, then filtering, then encoding or multiple encoders
[22:51:49 CEST] <JEEB> all of those generally can have N>=1 threads
[23:04:07 CEST] <Marshalleq> Thanks.  So other question, I'm running on two AMD systems TR 1950X and Ryzen 1850X - The Threadripper runs KVM/XEN Virtualisation and has various emulation modes that can be presented to the VM.  I haven't found a list of what CPU extensions ffmpeg uses, so that I can understand what performance hit might be present if running virtualised in e.g. Q35-3.1.  Is there a list anywhere?  
[00:00:00 CEST] --- Wed Jun 19 2019