kurosu has quit [Quit: Connection closed for inactivity]
noonien8 has quit [Quit: Ping timeout (120 seconds)]
noonien8 has joined #ffmpeg-devel
derpydoo has joined #ffmpeg-devel
kasper93_ has joined #ffmpeg-devel
kasper93 has quit [Ping timeout: 255 seconds]
thilo has quit [Ping timeout: 255 seconds]
thilo has joined #ffmpeg-devel
thilo has quit [Changing host]
thilo has joined #ffmpeg-devel
wangbin has joined #ffmpeg-devel
MisterMinister has joined #ffmpeg-devel
cone-652 has joined #ffmpeg-devel
<cone-652>
ffmpeg Andreas Rheinhardt release/6.1:5eca8964a932: fftools/ffmpeg_mux_init: Restrict disabling automatic copying of metadata
<cone-652>
ffmpeg Andreas Rheinhardt release/6.0:330b8305c1cd: fftools/ffmpeg_mux_init: Restrict disabling automatic copying of metadata
lemourin has quit [Read error: Connection reset by peer]
lemourin has joined #ffmpeg-devel
SuperFashi has quit [Quit: No Ping reply in 180 seconds.]
SuperFashi has joined #ffmpeg-devel
jamrial has quit []
derpydoo has quit [Ping timeout: 255 seconds]
aaabbb has joined #ffmpeg-devel
Xaldafax has quit [Quit: Bye...]
durandal_1707 has quit [Ping timeout: 272 seconds]
durandal_1707 has joined #ffmpeg-devel
cone-652 has quit [Quit: transmission timeout]
<Lynne>
aac is everywhere :(
<psykose>
it's like that "we've got you surrounded" meme
<aaabbb>
i posted about this in #ffmpeg but this might be a more appropriate place. i found and confirmed a bug in the latest version of ffmpeg, caused by encoding to flac with cholesky as the lpc algorithm using 11 or more passes, with compression_level 2 or lower with certain inputs (such as velvet noise, but not white noise etc). it doesn't happen if the sample format is set to s32
<mindfreeze>
I am thinking will this be superseded with native decoder if both are merged, or are they trying to have external decoder + native decoder?
<nevcairiel>
there can be more then one person trying to do different things :D
<JEEB>
mindfreeze: fraunhofer wants to push their thing while we have the ffvvc under review so we can internally provide it
<jdarnley>
Wow vpclmulqdq is rubbish. No faster to do fewer instructions on ymm rather than more on xmm.
<Gramner>
it's 1p5 regardless of xmm/ymm/zmm
Krowl has quit [Read error: Connection reset by peer]
dellas has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
<jdarnley>
Why are there 2 separate tables in intel's doc under CPUID detailing the values with conflicting information?
<jdarnley>
oh my bad they are for separate eax values
Jackarain has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
cone-290 has joined #ffmpeg-devel
<cone-290>
ffmpeg Zhao Zhili master:e920a8480185: mailmap: remap my email accounts
lexano has joined #ffmpeg-devel
dellas83 has joined #ffmpeg-devel
Jackarain has quit [Quit: Client closed]
dellas has quit [Ping timeout: 258 seconds]
Krowl has quit [Read error: Connection reset by peer]
MrZeus has joined #ffmpeg-devel
MrZeus_ has joined #ffmpeg-devel
dellas8386 has joined #ffmpeg-devel
dellas83 has quit [Read error: Connection reset by peer]
MrZeus has quit [Ping timeout: 260 seconds]
Krowl has joined #ffmpeg-devel
<Gramner>
jdarnley: oh the >128-bit clmul on ICL is actually emulated with 128-bit ops. it's only "properly" implemented since SPR (and zen 4)
<Gramner>
so it depends on what hw you're using
<jdarnley>
lol nice
<jdarnley>
in this case a Silver 4316
<Gramner>
yeah that's ice lake
<jdarnley>
the avx2 function is good ebough already twice the speed of the large LUT version we used previously
<cone-290>
ffmpeg Andreas Rheinhardt master:155f0c8ef743: avformat/webpenc: Check seeks
<jdarnley>
and we have an avenue for more optimization by mergeing it with something else to reduce reads and writes
<frankplow>
mindfreeze: I think they want both VVdeC and ffvvc. The reasoning they give in the ML is that ffvvc is not yet a fully-compliant VVC decoder as it is missing some features.
<mindfreeze>
frankplow: says every reference implementation;)
<mindfreeze>
I meant people who make reference implementation
<another|>
vvdec is not a reference impl
<mindfreeze>
Well it is from same team?
<mindfreeze>
I thought it was same people, i could be wrong
<frankplow>
mindfreeze: Yeah it's also from the Fraunhofer institute but it's a different application to the reference software VTM. vvdec is meant to be a production decoder I believe.
BetweenUs has joined #ffmpeg-devel
rene17 has joined #ffmpeg-devel
BetweenUs has quit [Ping timeout: 246 seconds]
<courmisch>
uh is CLMUL used for anything but crypto?
<Gramner>
it's used for crc calculations
<kierank>
I was going to talk about this at vdd
<kierank>
but hungover kieran is not going to be able to explain this in 10 mins
<Gramner>
the underlying math is very much non-obvious
<kierank>
yes and written like a true maths paper where the author says stuff is obvious
<kierank>
when it is not
<elenril>
it's not a true maths paper unless the author also insults your intelligence
<elenril>
"it must be obvious to any educated reader that..."
<Gramner>
fwiw I'm pretty sure you can do that main sdi crc loop in like 7 instructions with avx-512, but dunno how well it could run compared to avx2 on ICL since apparently 256-bit clmul sucks there
kasper93_ is now known as kasper93
<Gramner>
elenril: I thought that's what you write when you don't understand it yourself and therefore you don't want the audience asking questions about it since it would implicitly make the asker appear stupid
<elenril>
both are in wide use I think
<kierank>
Gramner: we are targeting old systems to be honest
<kierank>
as it's some legacy crc
qeed_ has quit [Quit: qeed_]
MrZeus__ has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
MrZeus_ has quit [Ping timeout: 260 seconds]
qeed has joined #ffmpeg-devel
Vedaa58193 has quit [Ping timeout: 255 seconds]
dellas8386 has quit [Read error: Connection reset by peer]
dellas8386 has joined #ffmpeg-devel
<courmisch>
what I don't get with CLMUL is that they never seem to tell what is the prime polynom
<courmisch>
I guess everybody uses the same one and it never seems to be explicit
Vedaa58193 has joined #ffmpeg-devel
<kierank>
yes
<kierank>
and the paper also assumes 8-bit data and mod-8 polynomial
<kierank>
crcs are weird, the notation is bitreversed, leading one, trailing one
<kierank>
it's like they talk their own language not explained to the world
AbleBacon has joined #ffmpeg-devel
<courmisch>
they could say GF(8) it's only 5 characters
blb has quit [Quit: brb]
blb has joined #ffmpeg-devel
microchip_ has quit [Quit: There is no spoon!]
microchip__ has joined #ffmpeg-devel
microchip__ is now known as microchip_
dellas8386 has quit [Remote host closed the connection]
dellas has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
feiw has quit [Ping timeout: 248 seconds]
feiw has joined #ffmpeg-devel
cone-290 has quit [Quit: transmission timeout]
beastd has joined #ffmpeg-devel
BetweenUs has joined #ffmpeg-devel
BetweenUs has quit [Read error: Connection reset by peer]
Krowl has joined #ffmpeg-devel
MrZeus_ has joined #ffmpeg-devel
MrZeus__ has quit [Ping timeout: 240 seconds]
<durandal_1707>
bug by design
<elenril>
huh, --disable-optimizations FATE takes only 30% longer than normal
<durandal_1707>
because -mmx removal
<elenril>
durandal_1707: do you care about sending commands to filters in ffmpeg CLI?
MrZeus__ has joined #ffmpeg-devel
<durandal_1707>
elenril: not really, there is better way to handle that, via sending commands using direct calls or other filters
MrZeus_ has quit [Ping timeout: 258 seconds]
<durandal_1707>
elenril: why? it looks trivial to keep it working.
<elenril>
not entirely with threading
<elenril>
and I don't know how to test it
dellas has joined #ffmpeg-devel
<durandal_1707>
you do not have a way to access random thread running filtergraph?
<elenril>
it is implemented
<elenril>
it's just not entirely trivial
<elenril>
and I don't know how to test it
<durandal_1707>
elenril: use -re and some filtergraph and stream video changing stuff in filters on the fly?
<elenril>
give example
<elenril>
stuff I tried fails with -38
<durandal_1707>
example?
Krowl has quit [Read error: Connection reset by peer]
<elenril>
what filter and what command
<durandal_1707>
the one that supports commands
<durandal_1707>
what exact stuff you tried?
navi has quit [Read error: Connection reset by peer]
<jamrial>
elenril: there's an example in ticket 10226
<kierank>
Is there an xmm/ymm transition penalty?
<jamrial>
don't think so. only the requirement to call vzeroupper at the end for using ymm regs
<Lynne>
there's a penalty for turning on the upper lane if it hasn't been used in a while IIRC, but could only be on old CPUs now
elastic_dog has quit [Ping timeout: 272 seconds]
Krowl has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
Krowl has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
elastic_dog has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
Mikhail_AMD has joined #ffmpeg-devel
dellas has quit [Remote host closed the connection]
<Lynne>
elenril: ping
cone-268 has joined #ffmpeg-devel
<cone-268>
ffmpeg Michael Niedermayer master:9450a4a7fef5: avfilter/framesync: fix order of operation with = and <0
<cone-268>
ffmpeg Michael Niedermayer master:c0a18e884c2d: avfilter/buffersink: fix order of operation with = and <0
<cone-268>
ffmpeg Michael Niedermayer master:03a4aa9699c3: avcodec/flicvideo: consider width in copy loops
<cone-268>
ffmpeg Michael Niedermayer master:4fb9d946883b: avformat/lafdec: Check for 0 parameters