michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 6.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
mkver has joined #ffmpeg-devel
lexano has quit [Ping timeout: 255 seconds]
sudden has quit [Server closed connection]
sudden has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
mkver has quit [Ping timeout: 258 seconds]
thilo has quit [Ping timeout: 252 seconds]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
cone-476 has quit [Quit: transmission timeout]
dellas has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
MrZeus_ has quit [Ping timeout: 255 seconds]
CAT_S has quit [Server closed connection]
CAT_S has joined #ffmpeg-devel
dellas has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
lemourin has quit [Read error: Connection reset by peer]
navi has quit [Quit: WeeChat 4.0.4]
lemourin has joined #ffmpeg-devel
jamrial has quit []
ldts has quit [Server closed connection]
ldts has joined #ffmpeg-devel
feiw1 has quit [Remote host closed the connection]
feiw1 has joined #ffmpeg-devel
durandal_1707 has quit [Ping timeout: 246 seconds]
durandal_1707 has joined #ffmpeg-devel
kurosu has quit [Quit: Connection closed for inactivity]
AbleBacon has quit [Read error: Connection reset by peer]
darkapex has joined #ffmpeg-devel
drv has quit [Server closed connection]
drv has joined #ffmpeg-devel
wangbin has joined #ffmpeg-devel
derpydoo has quit [Ping timeout: 260 seconds]
Traneptora has quit [Quit: Quit]
sepro2 is now known as sepro
<JEEB>
-_- time to find out where the heck in my email client the cover post of the latest version of the threading change set is
<elenril>
some individual patches are up to v3, but there's just one thread
cone-026 has joined #ffmpeg-devel
<cone-026>
ffmpeg Gyan Doshi master:67a2571a5547: avcodec/libsvtav1: add version guard for external param
<elenril>
I'll probably push the first 7 patches today and send a new set with the rest
<JEEB>
and my email client has gone completely bonkers esp. with some of the responses
<kurosu>
errr, I thought I posted an answer here about the scalar product ssse3
<kurosu>
courmisch: when ssse3 was introduced unaligned access was extremely expensive. It was much improved maybe 3 generations later (I wonder if anyone has concrete difference between using movdqa/aps and movdqu/ups on recent nodes). The reason for the ssse3 is palignr
<kurosu>
now, if and when an avx2 version exists, I'd say it boils down to whether to help older CPUs where this is a benefit
<kurosu>
dav1d made that choice but different constrains I guess
Krowl has joined #ffmpeg-devel
<kurosu>
Went through 2 renames: b1159ad92818cd8f0885d252b0800f5960fe7241
nevcairiel has quit [Server closed connection]
nevcairiel has joined #ffmpeg-devel
<kurosu>
so, not so much align as cacheline splits, but same analysis why 10+ years later it doesn't matter for an avx2-capable CPU
<Lynne>
kurosu: there's no difference between mova and movu on an aligned address these days
<Lynne>
otherwise, alignment penalty is in the noise
<kurosu>
Yes, my point
<Lynne>
I don't have anything concrete, though, I do remember some blog post which went through a lot of engineering to make alignment errors matter, and it's 8 years old by now
<kurosu>
These days, targetting olderer systems require tuning on them
<kurosu>
But yes. The commit msg says core2, so...
<kurosu>
More than 8y
<kurosu>
Reminds me of kierank's talk (which I hand no time to watch/track how to watch). People may scoff at this, but I don't know many companies writing "raw" assembly anymore. Intrinsics it is, if not a particular abstraction layer to target several archs at once. Rationale is it is costly/not a talent commodity (on the contrary, junior engineers may be the ones tasked with it) and the x% loss is not worth it in the grand scheme
<kurosu>
of things for these companies
<elenril>
mwahaha, I'll get a replacement CPU today
<courmisch>
this is audio stuff. unless you are transcoding many streams, you don't need optimisations at all
<elenril>
my desktop will finally live again, after a month
<courmisch>
and in many cases we'd get some speedup by adding restrict qualifiers where possible
<kierank>
kurosu: <3
<JEEB>
elenril: meanwhile I actually visited a physical electronics store yesterday to test the videotoolbox code I wrote
<JEEB>
curl and execute and hope no-one looks at you in a weird way
<Lynne>
an apple store?
<JEEB>
nah, we have zero of those
<durandal_1707>
abusing stores
<JEEB>
just a large store which happens to enable wifi and internets on the test machines and that's large enough that you won't get salesperson poking at you right away
<elenril>
doesn't kierank host some apple machines for devs?
<JEEB>
yea but M1
<JEEB>
this requires M3 hardware
<JEEB>
because no software decoder
<elenril>
ah
<JEEB>
even if dav1d is distributed as part of safari for avif
<elenril>
i suppose expecting apple to provide us with hw is a silly notion
<wbs>
elenril: lol, yes, that's a very silly notion
<courmisch>
yeah, why do we care to implement this if nobody provides hardware
Krowl has quit [Read error: Connection reset by peer]
<JEEB>
because on trac someone did a feature request and it looked like a small thing to poke at
<JEEB>
thankfully my branch does not work :P
<galad>
dav1d is distributed as part of macOS and iOS, but it's used only to decode avif images at the moment
<galad>
I couldn't find a way to load it to decode a video yet
<JEEB>
ah so it is not a part of safari only
<galad>
it's in /System/Library/Video/Plug-Ins/AV1DecoderSW.bundle
<courmisch>
RVV 1.0 boards are back in stock, FYI
<courmisch>
no excuse for using C906 protos any more
<kierank>
elenril: I would suggest getting an M3 but there is no Mac mini M3 afaik
<JEEB>
yea, only macbooks for now
<courmisch>
I wouldn't buy it anyway unless there's a sponsor
ccawley2011 has joined #ffmpeg-devel
<elenril>
>Your parcel is out for delivery and will be delivered today between 14:08-15:08
<elenril>
interesting time range
<courmisch>
the lunch break has been extended by 8 minutes and 21 seconds due to regulatory reasons
Krowl has joined #ffmpeg-devel
derpydoo has quit [Ping timeout: 240 seconds]
MrZeus_ has joined #ffmpeg-devel
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
elastic_dog has quit [Ping timeout: 258 seconds]
elastic_dog has joined #ffmpeg-devel
BetweenUs has joined #ffmpeg-devel
cone-026 has quit [Quit: transmission timeout]
<durandal_1707>
courmisch: you are attacking me on ML!
<kierank>
durandal_1707: probably deserved
<aaabbb>
which thread?
<durandal_1707>
kierank: your agressive behavior will not be tolerated here.
mkver has joined #ffmpeg-devel
<kierank>
lol
<courmisch>
durandal_1707: I said your message sounded p-a because it does
<courmisch>
I didn't attack *you*. I don't call human beings "it"
Traneptora has joined #ffmpeg-devel
<durandal_1707>
courmisch: you are attacking my reactions and emotions
<courmisch>
durandal_1707: pretty rich from somebody who called my patches "spam"
<durandal_1707>
courmisch: it was a joke
<courmisch>
you're allowed to be passive-aggressive and I'm allowed to call you out. Get on with it.
<durandal_1707>
courmisch: i was not passive-aggressive at all, just stated obvious fact that Ryzen is not available for everyone, and thus there is no point into removing SIMD code.
navi has joined #ffmpeg-devel
<courmisch>
durandal_1707: there's a big difference between saying that XX hardware is faster SSSE3 like kurosu did, and just stating that not everyone has a Ryzen
<courmisch>
and however you meant it, it *sounded* passive-aggressive to me. which is all I wrote
jamrial has joined #ffmpeg-devel
<durandal_1707>
sounded == attack
<jamrial>
wellsakus: could you address Neal Gompa's emails about evc?
JEEB has quit [Server closed connection]
JEEB has joined #ffmpeg-devel
BetweenUs has quit [Quit: Leaving]
ccawley2011 has quit [Read error: Connection reset by peer]
ccawley2011 has joined #ffmpeg-devel
ccawley2011 has quit [Read error: Connection reset by peer]
ccawley2011 has joined #ffmpeg-devel
<courmisch>
durandal_1707: no and false accusations = attacks
<Lynne>
why did parsers ever become a part of lavc rather than lavf?
<Lynne>
surely splitting up frames is best done in demuxers
<courmisch>
because multimedia is a pile of hacks
<courmisch>
:(
<durandal_1707>
courmisch: you started first
Krowl has quit [Read error: Connection reset by peer]
lexano has joined #ffmpeg-devel
<aaabbb>
is this joking banter or is it really this toxic?
<durandal_1707>
joke become toxic
dellas has joined #ffmpeg-devel
derpydoo has joined #ffmpeg-devel
<elenril>
Lynne: because the code is in lavc
<elenril>
and often shared with decoders
Krowl has joined #ffmpeg-devel
<courmisch>
durandal_1707: that your answer sounded a-p can't be a false accusations since it's a nonfalsifiable subjective perception. you're the one who started accusing me of attacking you. I'll leave it there because I'm not interested in your ramblings TBH
dellas83 has joined #ffmpeg-devel
dellas has quit [Read error: Connection reset by peer]
<durandal_1707>
TBH you are finding conflicts everywhere
<Traneptora>
I'm confused about the re-vote, is it possible to just ignore the declaration and move on?
<Traneptora>
and who has the authority to just say "ignore it and move on"
<Traneptora>
or is it going to start into another big email chain that goes nowhere
<elenril>
I've posted my unused link publicly on the ML
<elenril>
so did some a few other people
<elenril>
presumably, if enough people do that, the vote will be clearly invalid
dellas83 has quit [Remote host closed the connection]
<durandal_1707>
all folks that did that should be immediatelly stripped from GA rights to vote at all and be immediatelly banned from FFmpeg project.
<courmisch>
durandal_1707: yes.
<courmisch>
durandal_1707: I'm sorry that you take it so badly. My point was really just that you could have worded it more straightforwardly and unambiguously, or better not reply at all.
<courmisch>
what's the policy on Obviously Fake author names?
dellas has joined #ffmpeg-devel
<courmisch>
elenril: colour me pessimistic. Well I suppose I am naturally pessimistic-coloured, so that won't make any difference
tortoise has quit [Server closed connection]
tortoise has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
<typological>
basically, converting from `sub.srt` to `sub.ass` format is producing a rounding error in the timestamps, I checked `libavformat/srtdec` and it decoded the timestamps properly, in milliseconds but parts of hundred seconds at `libavformat/assenc.c` is not rounded properly, and I do not know where between these two the rounding happens, I would be
<typological>
happy if someone can point me at something that I can start with, any guesses are welcome.
elastic_dog has joined #ffmpeg-devel
sudden has quit [Ping timeout: 255 seconds]
sudden has joined #ffmpeg-devel
<mkver>
typological: 1. ASS uses only 10ms precision. 2. AVPackets store start time and duration and not end time; in this case, the first packet had a duration of 4766ms which gets rounded to 4770ms when converting to the ASS timebase. The start time also gets rounded upwards and this adds a +1 to the end time that would not have happened had the end time been rounded to 10ms itself.
Krowl has quit [Read error: Connection reset by peer]
___nick___ has joined #ffmpeg-devel
___nick___ has quit [Client Quit]
<typological>
I see, very helpful. thank you
___nick___ has joined #ffmpeg-devel
<typological>
what would your suggestion be to solve this? my first thought is to write a new function for avpacket that rounds does duration + start time, and round that, then subtract start time and make that the new duration.
<iive>
why round at all? there is a timestamp type that could be used for precise timing
<iive>
oh, the ass needs the rounding..
<typological>
yeah... very annoying that it uses 10ms precision
___nick___ has quit [Ping timeout: 246 seconds]
navi has quit [Quit: WeeChat 4.0.4]
navi has joined #ffmpeg-devel
HarshK23 has joined #ffmpeg-devel
kurosu has quit [Quit: Connection closed for inactivity]
dellas has quit [Remote host closed the connection]
dellas has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
galad has quit [Remote host closed the connection]