#ffmpeg-devel on 2025-04-10 — irc logs at libera.irclog.whitequark.org

2025-03-03 01:04 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

00:13 witchymary has joined #ffmpeg-devel

00:15 usagi_mimi has joined #ffmpeg-devel

00:19 thilo has quit [Ping timeout: 260 seconds]

00:20 thilo has joined #ffmpeg-devel

00:25 Kei_N_ has joined #ffmpeg-devel

00:28 Kei_N has quit [Ping timeout: 265 seconds]

00:39 Traneptora has joined #ffmpeg-devel

00:44 jdek has quit [Ping timeout: 246 seconds]

00:46 jdek has joined #ffmpeg-devel

00:51 \\Mr_C\\ has quit [Remote host closed the connection]

01:04 usagi_mimi has quit [Ping timeout: 248 seconds]

01:13 cone-397 has joined #ffmpeg-devel

01:13 <cone-397> ffmpeg Jerome Martinez master:3aec1f87c93a: avcodec/ffv1enc: permit 1024 slices

01:13 <cone-397> ffmpeg Michael Niedermayer master:ce538ef97a7b: swscale/output: Fix integer overflow in yuv2gbrp_full_X_c()

01:13 <cone-397> ffmpeg Michael Niedermayer master:fbf002dbd2a2: avcodec/pngenc: set default prediction method to PAETH

01:13 <cone-397> ffmpeg Michael Niedermayer master:c6a889f3e092: avcodec/osq: avoid undefined negation

01:13 <cone-397> ffmpeg Michael Niedermayer master:e47a3c7d62af: doc/encoders: Document compression_level for PNG

01:13 <cone-397> ffmpeg Michael Niedermayer master:d16a058dbc99: swscale/swscale: Do not crash on floats

01:13 <cone-397> ffmpeg Michael Niedermayer master:f6cd9bfa829e: doc/encoders: Document png pred

01:23 usagi_mimi has joined #ffmpeg-devel

01:49 abdu has quit [Quit: Client closed]

02:56 derpydoo has joined #ffmpeg-devel

03:21 jamrial has quit []

03:28 Anthony_ZO has joined #ffmpeg-devel

03:48 derpydoo has quit [Quit: derpydoo]

04:13 cone-397 has quit [Quit: transmission timeout]

04:55 <fflogger> [editedticket] JackLau: Ticket #11021 ([ffplay] ffplay with nodisp flag on RTSP input fails to open stream) updated https://trac.ffmpeg.org/ticket/11021#comment:1

05:06 <fflogger> [editedticket] madsfoto: Ticket #11271 ([undetermined] How to choose the best hwaccel?) updated https://trac.ffmpeg.org/ticket/11271#comment:6

05:14 paulk-bis has joined #ffmpeg-devel

05:16 paulk has quit [Ping timeout: 248 seconds]

05:42 NotWarcop has quit [Remote host closed the connection]

06:20 <fflogger> [editedticket] JackLau: Ticket #10913 ([undetermined] Generated sdp file missing audio) updated https://trac.ffmpeg.org/ticket/10913#comment:1

06:23 <fflogger> [editedticket] Balling: Ticket #11271 ([undetermined] How to choose the best hwaccel?) updated https://trac.ffmpeg.org/ticket/11271#comment:7

06:52 <fflogger> [editedticket] Balling: Ticket #11271 ([undetermined] How to choose the best hwaccel?) updated https://trac.ffmpeg.org/ticket/11271#comment:8

06:53 Anthony_ZO1 has joined #ffmpeg-devel

06:56 Anthony_ZO has quit [Ping timeout: 252 seconds]

06:56 Anthony_ZO1 is now known as Anthony_ZO

07:07 aaabbb has quit [Ping timeout: 265 seconds]

07:30 j45_ has joined #ffmpeg-devel

07:31 j45 has quit [Ping timeout: 260 seconds]

07:31 j45_ is now known as j45

07:31 j45 has quit [Changing host]

07:31 j45 has joined #ffmpeg-devel

07:34 mkver has joined #ffmpeg-devel

07:41 ngaullier has joined #ffmpeg-devel

07:42 ngaullier has quit [Remote host closed the connection]

07:46 ngaullier has joined #ffmpeg-devel

08:10 derpydoo has joined #ffmpeg-devel

08:18 ^Neo has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]

08:19 ^Neo has joined #ffmpeg-devel

08:19 ^Neo has quit [Changing host]

08:28 Martchus_ is now known as Martchus

08:35 paulk-bis has quit [Quit: WeeChat 3.0]

08:36 paulk has joined #ffmpeg-devel

08:36 paulk has quit [Changing host]

08:36 paulk has joined #ffmpeg-devel

08:40 mkver has quit [Remote host closed the connection]

08:40 mkver has joined #ffmpeg-devel

08:58 <haasn> it would be neat if checkasm could report a coverage summary for a platform, i.e. highlighting functions that are still C only

09:03 martinr1 has quit [Quit: WeeChat 4.5.2]

10:04 abdu has joined #ffmpeg-devel

10:08 psilokos has joined #ffmpeg-devel

10:10 abdu has quit [Quit: Client closed]

10:14 abdu has joined #ffmpeg-devel

10:24 hbbs has quit [Ping timeout: 252 seconds]

10:33 hbbs has joined #ffmpeg-devel

10:33 hbbs has quit [Changing host]

10:33 hbbs has joined #ffmpeg-devel

10:36 cone-531 has joined #ffmpeg-devel

10:36 <cone-531> ffmpeg Martin Storsjö master:4d4b301e4a26: checkasm: hevc_pel: Use helpers for checking for writes out of bounds

10:43 <haasn> ramiro: heads up: switched from 3/5/7 for the dither y offset to 2/3/5

10:43 <haasn> better approximates phi

10:44 <haasn> and also works better modulo 4

10:45 <haasn> (for 2x2, it aligns Y=V, U=A which is the best alignment; ditto R=B, G=A)

10:46 <haasn> err, other way around, Y=U, V=A; R=G, B=A

11:01 derpydoo has quit [Quit: derpydoo]

11:14 av500 has quit [Remote host closed the connection]

12:06 Guest34 has joined #ffmpeg-devel

12:08 <Traneptora> gotta have that fibonnaci phi approximation

12:08 <haasn> dither implemented in my x86 backend, now just needs linear and we're good to go :D

12:17 IndecisiveTurtle has quit [Remote host closed the connection]

12:18 jamrial has joined #ffmpeg-devel

12:40 IndecisiveTurtle has joined #ffmpeg-devel

12:47 abdu has quit [Ping timeout: 240 seconds]

12:56 abdu has joined #ffmpeg-devel

13:06 <fflogger> [newticket] sch: Ticket #11544 ([undetermined] Seex playback stutters) created https://trac.ffmpeg.org/ticket/11544

13:06 abdu has quit [Ping timeout: 240 seconds]

13:12 Guest34 has quit [Ping timeout: 240 seconds]

13:13 blb has quit [Ping timeout: 248 seconds]

13:14 blb has joined #ffmpeg-devel

13:30 BradleyS_ has joined #ffmpeg-devel

13:31 signalhunter5 has joined #ffmpeg-devel

13:32 cosminaught_ has joined #ffmpeg-devel

13:32 psykose_ has joined #ffmpeg-devel

13:32 sm2n_ has joined #ffmpeg-devel

13:32 kierank_ has joined #ffmpeg-devel

13:32 Taripe has joined #ffmpeg-devel

13:32 jkhsjdhjs_ has joined #ffmpeg-devel

13:35 welder_ has joined #ffmpeg-devel

13:36 cone-531 has quit [Quit: transmission timeout]

13:40 IndecisiveTurtle has quit [*.net *.split]

13:40 psykose has quit [*.net *.split]

13:40 BradleyS has quit [*.net *.split]

13:40 sm2n has quit [*.net *.split]

13:40 michaelni has quit [*.net *.split]

13:40 signalhunter has quit [*.net *.split]

13:40 welder has quit [*.net *.split]

13:40 ePirat has quit [*.net *.split]

13:40 jkhsjdhjs has quit [*.net *.split]

13:40 kierank has quit [*.net *.split]

13:40 cosminaught has quit [*.net *.split]

13:40 cosminaught_ is now known as cosminaught

13:40 kierank_ is now known as kierank

13:40 psykose_ is now known as psykose

13:40 BradleyS_ is now known as BradleyS

13:40 jkhsjdhjs_ is now known as jkhsjdhjs

13:40 sm2n_ is now known as sm2n

13:40 signalhunter5 is now known as signalhunter

13:42 IndecisiveTurtle has joined #ffmpeg-devel

13:43 abdu has joined #ffmpeg-devel

13:46 michaelni has joined #ffmpeg-devel

13:48 abdu has quit [Ping timeout: 240 seconds]

13:49 <mkver> Is there ever a reason to zero a register by loading a zero from memory instead of just xoring?

13:51 minimal has joined #ffmpeg-devel

13:57 <jamrial> doubt it

14:08 abdu has joined #ffmpeg-devel

14:12 abdu has quit [Ping timeout: 240 seconds]

14:15 <BBB> mkver: as a static (known) constant, no, but if you were to load a variable number (where 0 is one of the possible values you may want to load), then of course yes

14:16 <mkver> It is a known constant. See git grep b00 libpostproc/

14:24 abdu has joined #ffmpeg-devel

14:31 <BBB> then the answer is no

14:31 <BBB> good find

14:49 <Lynne> mkver: if you overload your ALUs and just have to reset a reg, what else would you do?

14:50 <Lynne> at 0.25 throughput and hundreds of instructions long pipeline, you can do whatever you want if you bottleneck that

14:51 <mkver> I don't think this was done with ALU overload in mind.

14:56 abdu has quit [Ping timeout: 240 seconds]

15:00 <jamrial> wbs: your x86_64-msvc2022-wine fate client is printing a bogus Compiler string in its report

15:01 <jamrial> the x86 one is fine

15:05 cone-433 has joined #ffmpeg-devel

15:05 <cone-433> ffmpeg James Almer master:a039726c2a09: avutil/x86/aes: remove a few branches

15:05 <cone-433> ffmpeg James Almer master:d79e4f6d0b7b: tests/fate/checkasm: add the aes test

15:13 aaabbb has joined #ffmpeg-devel

15:20 abdu has joined #ffmpeg-devel

15:27 abdu has quit [Ping timeout: 240 seconds]

15:30 System_Error has quit [Ping timeout: 264 seconds]

15:34 <wbs> jamrial: it occasionally does that, it's usually good on the next run

15:39 abdu has joined #ffmpeg-devel

15:49 System_Error has joined #ffmpeg-devel

16:04 abdu has quit [Ping timeout: 240 seconds]

16:04 abdu99 has joined #ffmpeg-devel

16:09 j-b has quit [Ping timeout: 272 seconds]

16:10 j-b has joined #ffmpeg-devel

16:15 abdu99 has quit [Quit: Client closed]

16:19 Anthony_ZO has quit [Ping timeout: 272 seconds]

16:24 abdu has joined #ffmpeg-devel

16:51 Traneptora has quit [Quit: Quit]

17:01 mkver has quit [Ping timeout: 260 seconds]

17:02 abdu has quit [Ping timeout: 240 seconds]

17:11 abdu has joined #ffmpeg-devel

17:12 ngaullier has quit [Remote host closed the connection]

17:18 abdu has quit [Quit: Client closed]

17:33 System_Error has quit [Ping timeout: 264 seconds]

17:37 mkver has joined #ffmpeg-devel

17:50 System_Error has joined #ffmpeg-devel

18:00 <Lynne> does anyone have access to weird systems with vulkan?

18:00 <Lynne> I define weird as anything except amd and nvidia on linux, so windows qualifies

18:05 cone-433 has quit [Quit: transmission timeout]

18:08 <Lynne> I need folks to test if https://github.com/cyanreg/intel_bda_test works fine, so we know which drivers to disable for the ffv1/vc2 vulkan code in lavc

18:34 Traneptora has joined #ffmpeg-devel

19:04 <Lynne> I got reports already that the code runs on rockchip and adreno GPUs, which is surprising

19:04 <Lynne> anyone with a windows amd or intel machine able to test the program?

19:05 abdu has joined #ffmpeg-devel

19:05 <Lynne> would be interesting to know if it runs on rpi4/5 machines too

19:12 compnn has quit [Remote host closed the connection]

19:13 Compn has joined #ffmpeg-devel

19:14 paulk has quit [Ping timeout: 268 seconds]

19:16 paulk has joined #ffmpeg-devel

19:26 <kepstin> here's what i get on my work machine (intel pre-xe integrated graphics, fedora 41 w/ffmpeg 7.1.1): https://gist.github.com/kepstin/20be0681648ba2703271ed829719956d

19:26 <kepstin> not sure exactly what i was expecting there.

19:26 <kepstin> i might be able to try a few windows machines but probably not today.

19:30 <linkmauve> Lynne, on a rk3588 using panfrost from Mesa master built in February, I get this: https://linkmauve.fr/files/rk3588-rc_vk_test.txt

19:31 <Lynne> thanks

19:31 <Lynne> kepstin: yeah, so far, intel on linux seem to be the only ones who fail the test

19:31 <Lynne> I've asked a driver dev to take a look at it, but it may take a few days

19:38 <jamrial> Lynne: ../vulkan.h:114:5: error: unknown type name 'pthread_mutex_t'

19:38 <jamrial> on windows, msys2 with mingw-w64

19:39 <jamrial> included pthread.g and it compiled

19:39 <jamrial> and the test seemingly passed http://pastie.org/p/3itYKnuQ9lDtyvsoCMd9OS/raw

19:41 <Lynne> thanks, fixed the windows issues

19:41 <Lynne> yeah, no surprises about nvidia passing, their drivers and implementations have been faultless so far

19:42 <jamrial> will see if i can try intel and amd on windows later

19:42 <Lynne> thanks, those are what I'm most interested in

19:43 <jannau> Lynne: asahi/linux https://paste.centos.org/view/2bb45d6e

19:46 <Lynne> amazing, even object buffer is supported on asahi, thanks

19:48 <linkmauve> I should test a 16k page kernel someday.

20:01 user23 has joined #ffmpeg-devel

20:18 abdu has quit [Ping timeout: 240 seconds]

20:33 abdu has joined #ffmpeg-devel

20:53 System_Error has quit [Remote host closed the connection]

21:00 System_Error has joined #ffmpeg-devel

21:09 abdu has quit [Quit: Client closed]

22:03 abdu has joined #ffmpeg-devel

22:43 mkver has quit [Quit: Leaving]

23:42 abdu has quit [Ping timeout: 240 seconds]

23:46 System_Error has quit [Ping timeout: 264 seconds]