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 7.1.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
cone-853 has joined #ffmpeg-devel
<cone-853>
ffmpeg Andreas Rheinhardt master:004cc60f0e30: avutil/avassert: Add av_unreachable() and av_assume() macros
<cone-853>
ffmpeg Andreas Rheinhardt master:fc9e2a92c12f: avcodec/amrwbdec: Mark default switch as unreachable
<cone-853>
ffmpeg Andreas Rheinhardt master:b5824a6dab6c: avcodec/proresenc_anatoliy: Mark impossible case as unreachable
<cone-853>
ffmpeg Andreas Rheinhardt master:7a27357636ff: avcodec/mpeg4videodec: Mark impossible switch case as unreachable
<cone-853>
ffmpeg Andreas Rheinhardt master:fc520ccb07d5: avcodec/pcm-dvdenc: Mark unreachable default cases as unreachable
<cone-853>
ffmpeg Andreas Rheinhardt master:22e6e4e914d2: avcodec/utvideoenc: Remove always-false pixel format check
<cone-853>
ffmpeg Andreas Rheinhardt master:d7d1379d06e5: avcodec/vlc: Make code more readable with av_unreachable()
<cone-853>
ffmpeg Andreas Rheinhardt master:988e319b406b: avcodec/dolby_e_parse: Use av_unreachable instead of av_assert0(0)
<cone-853>
ffmpeg Andreas Rheinhardt master:4484e9b37307: avcodec/put_bits: Allow to mark places where PutBitContext is flushed
<cone-853>
ffmpeg Andreas Rheinhardt master:2df9c8da72d5: avcodec/e?ac3enc: Inform compiler about PutBitContext being blank
<cone-853>
ffmpeg Andreas Rheinhardt master:831a0d0c6bc1: avcodec/speedhqenc: Use av_unreachable() for unreachable condition
<cone-853>
ffmpeg Andreas Rheinhardt master:57696f7c2b42: avcodec/wmaenc: Use av_unreachable() instead of av_assert0(0)
<cone-853>
ffmpeg Andreas Rheinhardt master:ef58de7fdc4d: avcodec/mpegvideo_{dec,motion}: Mark unreachable code as unreachable
<cone-853>
ffmpeg Andreas Rheinhardt master:5d0e8fa4bb12: avcodec/vlc: Make assert check more strict
<cone-853>
ffmpeg Andreas Rheinhardt master:a35917268dfe: avcodec/mpegvideo encs: Add put_bits_assume_flushed() to encode_header
<cone-853>
ffmpeg Andreas Rheinhardt master:2282a3740ba5: avcodec/rv20enc: Use av_assert1() instead of av_assert0()
<cone-853>
ffmpeg Andreas Rheinhardt master:0bdb69119aaa: avcodec/vp9: Replace av_assert(0) by av_unreachable()
<cone-853>
ffmpeg Andreas Rheinhardt master:85fb3453f892: avcodec/adpcm: Use av_unreachable() instead of av_assert0()
<cone-853>
ffmpeg Andreas Rheinhardt master:5653e0b1fa20: avcodec/mjpegenc_common: Use av_unreachable() instead of av_assert0(0)
<cone-853>
ffmpeg Andreas Rheinhardt master:328ba54efa3b: avcodec/4xm: Use av_unreachable() instead of av_assert0(0)
<cone-853>
ffmpeg Andreas Rheinhardt master:2a5ee83442fd: avcodec/atrac3: Use av_unreachable() instead of av_assert1(0)
omegatron has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
Traneptora_ has quit [Quit: Quit]
Traneptora has joined #ffmpeg-devel
<haasn>
ramiro: maybe we should have SWS_OP_SWAP_BYTES not affect SwsComps.min/max? since it isn't really changing the value range so much as changing the representation
<haasn>
I mean we can keep track of a `bool swapped` next to the min/max if we really need to distinguish between them for some reason
<haasn>
alternatively we can move a `bool swapped` to the SwsOpRead/SwsOpWrite
<haasn>
to avoid the issue altogether
<ramiro>
haasn: good ideas
<ramiro>
it always did seem odd to have the byteswapped value for 12-bit being almost a full 16-bit
minimal has joined #ffmpeg-devel
MisterMinister has quit [Ping timeout: 252 seconds]
<mkver>
haasn: libswscale/x86/ops.c:507:1: warning: initializer overrides prior initialization of this subobject [-Winitializer-overrides]
<mkver>
Furthermore, static flexible array member initialization is not allowed in ISO C.
<mkver>
You are also using an insane amount of .data.rel.ro.
ccawley2011 has joined #ffmpeg-devel
<Lynne>
haasn: do you want to resubmit the 2 new vulkan filters you sent with USAGE_BUFFER_ADDRESS?
<Lynne>
if not, feel free to push them if you fixed them, they looked fine
<toots5446>
michaelni: thanks again for catching this test failure. Do you recommend I push a new version on the ML or a branch and link it here?
<ePirat>
toots5446, probably push a new version to the ML
ccawley2011_ has joined #ffmpeg-devel
ccawley2011 has quit [Ping timeout: 260 seconds]
<jamrial>
Gramner: no avx2 love? :p
<Gramner>
get a better cpu!
<jamrial>
i have an alder lake!
<jamrial>
Intel sucks for dropping avx512 :/
Traneptora has joined #ffmpeg-devel
<Gramner>
yeah, intel has a weird history of screwing around with ISA features. they did backtrack on the silly avx10 nonsense though, so at least future cpus will have a more sane feature set
<jamrial>
what ever happened with avx10?
<jamrial>
also, my cpu does have avx-vnni, so that's something
<Gramner>
they ditched the concept of 256-bit avx10. it will all be 512-bit
<Gramner>
not sure which gen will introduce it on client side. haven't really kept up with intel client µarch stuff after they stopped having avx-512
ccawley2011_ has quit [Ping timeout: 265 seconds]
Everything has joined #ffmpeg-devel
<kierank>
Gramner: can you make some numbers vs c for your latest patch
<kierank>
so I can make twitter lose its mind
<kierank>
oh you are comparing against c
<kierank>
I see
<Gramner>
not crazy amazing numbers vs c for those functions due to all being in 32-bit precision
<kierank>
Gramner: fyi "not that i doubt that compiler is not generating good assembly but all these instructions can be used in c as well if enable some compilation flags."
<kierank>
this guy at least has a semi reasonable point
iive has joined #ffmpeg-devel
<fflogger>
[newticket] Va1DER: Ticket #11602 ([undetermined] Muxing MKV and adding external SRT as anything except the only subtitle stream in an MKV prevents subtitle display in VLC) created https://trac.ffmpeg.org/ticket/11602
<haasn>
Lynne: No, see my commit for automatically adding this usage flag
<haasn>
I think the filter should only declare flags it needs itself
<fflogger>
[editedticket] Va1DER: Ticket #11602 ([undetermined] Muxing MKV and adding external SRT as anything except the only subtitle stream in an MKV prevents subtitle display in VLC) updated https://trac.ffmpeg.org/ticket/11602#comment:1
<haasn>
kierank: I tried with swscale, and it turns out the compiler just really, really hates using a lot of available instructions unless you A) use git master gcc, B) compile some functions with gcc and others with clang, C) jump through hoops to mangle your logic to babysit the compiler, and D) use intrinsics
<haasn>
and end up with uglier and less readable code than raw asm
kasper93 has quit [Quit: kasper93]
kasper93 has joined #ffmpeg-devel
mkver has quit [Ping timeout: 252 seconds]
cone-780 has quit [Quit: transmission timeout]
sepro has quit [Quit: Bye!]
sepro has joined #ffmpeg-devel
BradleyS has quit [Read error: Connection reset by peer]