[Ffmpeg-devel-irc] ffmpeg-devel.log.20190518

burek burek021 at gmail.com
Sun May 19 03:05:03 EEST 2019


[00:12:20 CEST] <BBB> Lynne: I think part of the problem is that the fft is generally not really used for cpu-constrained tasks. fft is typically audio-only anyway, and speed is mostly an issue for video (not audio) en (not de) coding, so ... it's not that it can't be done, it's just for lack of RoI
[00:12:33 CEST] <BBB> (but I think it' sgreat that you're doing this)
[00:13:27 CEST] <BBB> brb
[00:14:32 CEST] <cehoyos> +1
[00:23:58 CEST] <fumoboy007> Hi, it seems the password reset functionality of patchwork.ffmpeg.org is not working; I never received any email. Who do I contact?
[01:36:33 CEST] <Lynne> btw there's literally nothing left in the codebase that's viable and useful to SIMD but this
[01:41:24 CEST] <BBB> there's nothing viable to simd but fft?
[01:45:51 CEST] <kierank> there's idct
[01:45:56 CEST] <kierank> the higher bit depth variants
[01:46:53 CEST] <Lynne> yes, they're all transforms and they can all fit in tx.c
[01:51:57 CEST] <Lynne> but there's nothing codec-specific to SIMD on popular codecs, its all been done
[01:57:19 CEST] <BBB> I honestly don't think that's true
[01:57:25 CEST] <BBB> tons of vp9 has only xmm
[01:57:38 CEST] <BBB> avx2 for idcts would be beneficial, and 8bit vp9 is quite popular
[02:03:40 CEST] <Lynne> isn't the only thing left for vp9 avx2 loopfilter?
[02:07:05 CEST] <BBB> inverse transforms, no?
[08:18:19 CEST] <grosso> hi
[08:20:55 CEST] <grosso> I'm looking for a way to send control messages through an rtmp streaming, in addition to audio and video
[08:45:28 CEST] <grosso> hi
[14:31:59 CEST] <whitestone> hi people
[14:32:23 CEST] <whitestone> has someone some knowhow about blackmagic and decklink?
[15:07:25 CEST] <durandal_1707> "I'm sorry if I'm Spamming you, please don't sue me!" - Just a mere Teenager, too young to abide by the rules.
[16:06:04 CEST] <whitestone> someone know this error
[16:06:06 CEST] <whitestone> There are not enough buffered video frames. Video may misbehave!
[17:58:55 CEST] <antonin> Hi I have submit a patch and have no response. This is a very small and atomic patch. Can someone told what mistake did I do ?: http://ffmpeg.org/pipermail/ffmpeg-devel/2019-May/243865.html
[17:59:13 CEST] <antonin> Thanks in advance
[17:59:32 CEST] <kierank> antonin: nothing, I have been meaning to apply it
[18:00:55 CEST] <antonin> Oh great ! Thank you ! 
[18:00:59 CEST] <Lynne> shouldn't the body of the loop be indented
[18:01:19 CEST] <kierank> true
[18:01:28 CEST] <kierank> antonin: you need to indent the loop
[18:01:56 CEST] <kierank> Do we accept (for int=0, now?
[18:03:01 CEST] <dastan1910> how i apply a patch on ffmeg?
[18:04:05 CEST] <kierank> dastan1910: you send the patch to ffmpeg-devel
[18:04:56 CEST] <antonin> ok I will indent the loop. J=0 is for the number of timecode values in SEI
[18:06:13 CEST] <kierank> antonin: yes, but I was asking about the C99 declaration in the loop?
[18:06:27 CEST] <kierank> not sure this is accepted now or not
[18:06:49 CEST] <kierank> it is allowed it seems
[18:06:58 CEST] <kierank> so yes you need to indent and i think patch is ok
[18:07:27 CEST] <antonin> ok thanks
[18:13:21 CEST] <mkver> Loops with variable declaration are explicitly allowed. See https://www.ffmpeg.org/developer.html#C-language-features
[00:00:00 CEST] --- Sun May 19 2019


More information about the Ffmpeg-devel-irc mailing list