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 7.0 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
iive has quit [Quit: They came for me...]
thilo has quit [Ping timeout: 272 seconds]
thilo has joined #ffmpeg-devel
\\Mr_C\\ has quit [Remote host closed the connection]
<rcombs>
JEEB: so, yes, packet pos was already correct for DVHS (but when we then tried to actually read a packet after seeking to the offset, we'd immediately desync since we didn't skip the timecode), but was incorrect for FEC (and also had the same symptom of desyncing)
<rcombs>
there are really 4 separate cases here: DVHE+from-start, DVHE+seek, FEC+from-start, and FEC+seek
<rcombs>
DVHE+from-start desynced immediately and could fail to resync if the timecode contained 0x47 (because we didn't set the timecode), and set correct pos values; DVHE+seek would desync for the same reason
<rcombs>
FEC+from-start already largely worked, but would incorrect packet pos values; FEC+seek would fail because the pos values were wrong
<rcombs>
so this fixes all of those: pos is always correct, and we always skip the extra bytes (whether they're at the start or at the end)
<rcombs>
raw_packet_size is never actually modified by this patch
<rcombs>
it's just used differently
Mister_D has quit [Ping timeout: 272 seconds]
MisterMinister has joined #ffmpeg-devel
sadome has quit [Remote host closed the connection]
<sfan5>
huh looks like the ML just doesn't deliver to outlook anymore
<JEEB>
or it does, but outlook barfs at it :s
<JEEB>
hotmail/outlook/whatever MS thing has been picky as heck during the years. although mkver is using it atm
<thardin>
TIL progressive segmented frames (PsF)
<mkver>
thardin: ?
<mkver>
JEEB: Last email I received was Lynne's message about pushing the arm aacdec optimizations.
<JEEB>
yea, just noted that you were seemingly OK with the ML as an outlook user
<thardin>
mkver: it's used for transporting progressive essence over interlaced stuff
<thardin>
I bet there's a way to carry it in MXF. it counts as MIXED_FIELDS for sure
rvalue has quit [Ping timeout: 268 seconds]
<Lynne>
mkver: the ML's certificate was updated around 12h ago
Sean_McG has joined #ffmpeg-devel
<Lynne>
that particular email I sent just to make sure it works, but I did it just before
* Sean_McG
peeks in
<Lynne>
looks like you may have been affected
<Lynne>
the old certificates we had for the ML's address were 9 years old, self-signed, were about to expire, and were for the wrong domain anyway
<Lynne>
how they worked is a mystery until I ran into issues with verification when I switched to self-hosting my mail
<Lynne>
they were updated just after the mail, and they are actually correct now, so if anything, it shouldn't be more broken
<thardin>
"Brought to life by the movie industry in the end of the 1990s, the original PsF specification was focused on 24 frame/s content resulting in existing interlaced equipment having to be modified for 48 Hz scanning rate in order to work properly with 24 frame/s content." aaaaaaaaaaaaa
<Sean_McG>
is this good or bad?
<thardin>
I bet there's 47.95 Hz tapes around
<thardin>
"cursed" as the kids say
rvalue has joined #ffmpeg-devel
<Sean_McG>
hmmm gcc 14.1 has been out for a bit. I should build it on my PPC FATE node -- takes 2.5 days though even if I restrict it to C/C++
MetaNova has quit [Quit: quit]
MetaNova has joined #ffmpeg-devel
<thardin>
will have a go at slice threading speedhqdec
<thardin>
I may or may not split it up into two stages (VLC decode followed by IDCT)
<mkver>
thardin: Will this make single-threaded or frame-threaded decoding slower?
<cone-112>
ffmpeg Andreas Rheinhardt master:4df43df6e14d: avcodec/bsf/dts2pts: Fix shadowing
<cone-112>
ffmpeg Andreas Rheinhardt master:06d1840b5654: configure, avcodec/Makefile: Add h264parse->h264data,golomb dependencies
<cone-112>
ffmpeg Andreas Rheinhardt master:b7b1a4e6c3ee: configure: Add missing ftr_parser->adts_header,mpeg4audio dependencies
Krowl has quit [Read error: Connection reset by peer]
<thardin>
mkver: two stage decode might, due to poorer cache coherence. hence why I might not do it
<thardin>
one stage is easier to implement so I'll probably do that first. should have the same performance as before
<thardin>
since decoded dct coeffs easily fit in cache
SystemError has quit [Remote host closed the connection]
<thardin>
one thing I might consider is being "smart" and switching between one stage and two stage if given enough threads. but probably better to have an option rather than magic
<cone-112>
ffmpeg Ramiro Polla master:27f6211c74f3: lavc/aarch64: fix include for cpu.h
<sfan5>
JEEB: subscription email arrived an hour late or so :)
<JEEB>
vOv
q66 has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
paulk has quit [Ping timeout: 256 seconds]
paulk has joined #ffmpeg-devel
* sdc
I have an issue where checkasm reports a segfault only on windows and if I call the function manually inside the checkasm func (ie not using call_new() ) I can't find the segfault, so I'm a bit stuck on how to proceed
<sdc>
I have an issue where checkasm reports a segfault only on windows and if I call the function manually inside the checkasm func (ie not using call_new() ) I can't find the segfault, so I'm a bit stuck on how to proceed
Krowl has quit [Read error: Connection reset by peer]
<pengvado>
sdc: it could be something like using a 64bit register where the function prototype only has int32. calling manually may zero-extend or sign-extend the input, whereas call_new() intentionally puts garbage in any registers that aren't guaranteed to be meaningful.
<pengvado>
ofc that's just one guess, but it's something that call_new() intentionally does differently.
<mkver>
sdc: What function?
Livio has joined #ffmpeg-devel
novaphoenix has quit [Quit: i quit]
novaphoenix has joined #ffmpeg-devel
blb has quit [Ping timeout: 260 seconds]
blb has joined #ffmpeg-devel
<thardin>
either speedhqenc or speedhqdec seem to not handle borders right at the moment
Krowl has joined #ffmpeg-devel
<thardin>
yep, there's no checking width % 16 == 8 in mpegvideo_enc.c
<thardin>
and if I disable the check for decode_speedhq_border()'s return value I get the same kind of broken output as the slice threaded version of the decoder I'm working on
<thardin>
the latter doesn't error out because execute2() sucks at handling job errors
<courmisch>
jamrial: seems the same as lpc32 but outputting 64-bit values
<jamrial>
yes
<courmisch>
that function is quite nasty, as you can't parallelise output
<jamrial>
yeah, which is why the sse4 version of lpc32 is barely faster than plain c
<cone-112>
ffmpeg James Almer master:5ba6f4e63e49: checkasm/flacdsp: add a test for wasted33
<cone-112>
ffmpeg James Almer master:eb5733f38f56: x86/flacdsp: add an SSE4 version of wasted33
novaphoenix has joined #ffmpeg-devel
<cone-112>
ffmpeg James Almer master:b450c44630ba: x86/flacdsp: remove unused parameters to pmacsdql macro
T2300 has quit [Read error: Connection reset by peer]
<courmisch>
jamrial: yeah so I can't do lpc33 on with my 128-bit vector hardware
<courmisch>
I only have 4 kilobits of vector state, and that seems to need 5
<Lynne>
you could try spilling to stack, but isn't your new board arriving soon?
<courmisch>
Carrier note: Depart from transit country or district 2024-05-12 19:04:00 GMT+2
<courmisch>
so it seems to be at some Western European air cargo terminal
<courmisch>
or it was 23 hours ago
paulk has quit [Quit: WeeChat 3.0]
j45_ has joined #ffmpeg-devel
j45 has quit [Ping timeout: 260 seconds]
j45_ is now known as j45
j45 has joined #ffmpeg-devel
j45 has quit [Changing host]
<mkver>
wbs: Would you think it acceptable if av_sdp_create() were to be changed to return AVERROR_BUFFER_TOO_SMALL in case the provided buffer is too small?
paulk has joined #ffmpeg-devel
paulk has quit [Changing host]
paulk has joined #ffmpeg-devel
Traneptora has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
kasper93 has quit [Ping timeout: 240 seconds]
kasper93 has joined #ffmpeg-devel
<wbs>
mkver: sounds reasonable
AbleBacon has joined #ffmpeg-devel
kasper93 has quit [Ping timeout: 246 seconds]
kasper93 has joined #ffmpeg-devel
<Lynne>
jkqxz: ping about the 8 to 10 conversion trigger
<Lynne>
radv enable vulkan decoding by default now
<another|>
Lynne: in mesa master?
ngaullier has quit [Ping timeout: 252 seconds]
<jkqxz>
Does it still happen after e28195bf4b0e? (See change to rvcn_dec_message_(hevc|av1)_t.)
IndecisiveTurtle has joined #ffmpeg-devel
<Lynne>
jkqxz: I don't think so, tried the usual replication cases and couldn't find replicate
<Lynne>
surely it hadn't been this simple all this time?
<jkqxz>
What, that it was exactly as I said and you needed to add some zeroes at the end of the decode message?
<wbs>
not sure if the proper fix is just to add suitable casts somewhere, or if we'd need to do something else to ensure that the IDirect3DDevice9 we get returned from IDirect3DDeviceManager9_LockDevice actually is a IDirect3DDevice9Ex, which we need for later operations, it seems
ccawley2011 has joined #ffmpeg-devel
kasper93 has quit [Ping timeout: 252 seconds]
<sdc>
pengvado: seems like the sign extension issue was it, thank you!
MikhailAMD has joined #ffmpeg-devel
MrZeus has joined #ffmpeg-devel
ccawley2011 has quit [Read error: Connection reset by peer]
MrZeus_ has joined #ffmpeg-devel
MrZeus has quit [Ping timeout: 260 seconds]
sadome has joined #ffmpeg-devel
sadome has quit [Excess Flood]
sadome has joined #ffmpeg-devel
sadome has joined #ffmpeg-devel
sadome has quit [Changing host]
sadome has quit [Remote host closed the connection]
sadome has joined #ffmpeg-devel
sadome has joined #ffmpeg-devel
sadome has quit [Changing host]
MisterMinister has joined #ffmpeg-devel
sadome has quit [Remote host closed the connection]
sadome has joined #ffmpeg-devel
sadome has joined #ffmpeg-devel
sadome has quit [Changing host]
sadome has quit [Excess Flood]
sadome has joined #ffmpeg-devel
sadome has quit [Remote host closed the connection]
sadome has joined #ffmpeg-devel
sadome has quit [Remote host closed the connection]
MisterMinister has quit [Remote host closed the connection]
jamrial has quit [Read error: Connection reset by peer]
jamrial has joined #ffmpeg-devel
AbleBacon has quit [Read error: Connection reset by peer]
AbleBacon has joined #ffmpeg-devel
sadome has joined #ffmpeg-devel
sadome has joined #ffmpeg-devel
sadome has quit [Excess Flood]
kasper93 has joined #ffmpeg-devel
iive has joined #ffmpeg-devel
Traneptora has quit [Quit: Quit]
Traneptora has joined #ffmpeg-devel
SystemError has quit [Remote host closed the connection]
SystemError has joined #ffmpeg-devel
mkver has quit [Ping timeout: 268 seconds]
psykose has joined #ffmpeg-devel
SystemError has quit [Ping timeout: 260 seconds]
darkapex has quit [Remote host closed the connection]
SystemError has joined #ffmpeg-devel
darkapex has joined #ffmpeg-devel
IndecisiveTurtle has quit [Ping timeout: 268 seconds]