Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Michael Niedermayer]

[-- Attachment #1.1: Type: text/plain, Size: 1266 bytes --]

On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> 
> fftools/ffmpeg: convert to a threaded architecture
> 
> Change the main loop and every component (demuxers, decoders, filters,
> encoders, muxers) to use the previously added transcode scheduler. Every
> instance of every such component was already running in a separate
> thread, but now they can actually run in parallel.
> 
> Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> JEEB to be more correct and deterministic.
> 
> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> ---

infinite loops this: (most of the time)

ffmpeg -i tickets///2531/interlaced_top.wmv -t 10 -bitexact -vcodec ffv1 -level 0 -y  file..avi

https://samples.ffmpeg.org/ffmpeg-bugs/trac/ticket2531/interlaced_top.wmv

thx

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Complexity theory is the science of finding the exact solution to an
approximation. Benchmarking OTOH is finding an approximation of the exact

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Michael Niedermayer]

[-- Attachment #1.1: Type: text/plain, Size: 1592 bytes --]

On Thu, Dec 14, 2023 at 12:16:14AM +0100, Michael Niedermayer wrote:
> On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> > ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> > 
> > fftools/ffmpeg: convert to a threaded architecture
> > 
> > Change the main loop and every component (demuxers, decoders, filters,
> > encoders, muxers) to use the previously added transcode scheduler. Every
> > instance of every such component was already running in a separate
> > thread, but now they can actually run in parallel.
> > 
> > Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> > JEEB to be more correct and deterministic.
> > 
> > > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> > ---
> 
> infinite loops this: (most of the time)
> 
> ffmpeg -i tickets///2531/interlaced_top.wmv -t 10 -bitexact -vcodec ffv1 -level 0 -y  file..avi
> 
> https://samples.ffmpeg.org/ffmpeg-bugs/trac/ticket2531/interlaced_top.wmv

also infinite loops rtmp using old rtmplite/rtmp.py as server
and
ffmpeg -bitexact -rtmp_app myapp -rtmp_live 0 -i rtmp://127.0.0.1/try -bitexact -t 5 file-sp.avi

thx

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

The real ebay dictionary, page 2
"100% positive feedback" - "All either got their money back or didnt complain"
"Best seller ever, very honest" - "Seller refunded buyer after failed scam"

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Michael Niedermayer]

[-- Attachment #1.1: Type: text/plain, Size: 1672 bytes --]

On Thu, Dec 14, 2023 at 12:16:14AM +0100, Michael Niedermayer wrote:
> On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> > ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> > 
> > fftools/ffmpeg: convert to a threaded architecture
> > 
> > Change the main loop and every component (demuxers, decoders, filters,
> > encoders, muxers) to use the previously added transcode scheduler. Every
> > instance of every such component was already running in a separate
> > thread, but now they can actually run in parallel.
> > 
> > Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> > JEEB to be more correct and deterministic.
> > 
> > > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> > ---
> 
> infinite loops this: (most of the time)
> 
> ffmpeg -i tickets///2531/interlaced_top.wmv -t 10 -bitexact -vcodec ffv1 -level 0 -y  file..avi
> 
> https://samples.ffmpeg.org/ffmpeg-bugs/trac/ticket2531/interlaced_top.wmv

also a related question
is there a way to make things execute serially in a predictive manner ?

For example running this twice
./ffmpeg -threads 1 -v repeat+verbose -nostats -y -report -i mm-short.mpg -bitexact -t 1 /tmp/file...avi
and cleaning addresses out with sed 's/@ 0x[0-9a-f]*//g'

produces 2 reports that still differ by ordering significantly

thx

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Democracy is the form of government in which you can choose your dictator

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Anton Khirnov]

Quoting Michael Niedermayer (2023-12-14 00:16:14)
> On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> > ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> > 
> > fftools/ffmpeg: convert to a threaded architecture
> > 
> > Change the main loop and every component (demuxers, decoders, filters,
> > encoders, muxers) to use the previously added transcode scheduler. Every
> > instance of every such component was already running in a separate
> > thread, but now they can actually run in parallel.
> > 
> > Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> > JEEB to be more correct and deterministic.
> > 
> > > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> > ---
> 
> infinite loops this: (most of the time)
> 
> ffmpeg -i tickets///2531/interlaced_top.wmv -t 10 -bitexact -vcodec ffv1 -level 0 -y  file..avi
> 
> https://samples.ffmpeg.org/ffmpeg-bugs/trac/ticket2531/interlaced_top.wmv

Did you test this and the other loop with 7cc4b306eb? I cannot reproduce
it here.

-- 
Anton Khirnov
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Anton Khirnov]

Quoting Michael Niedermayer (2023-12-14 03:07:13)
> On Thu, Dec 14, 2023 at 12:16:14AM +0100, Michael Niedermayer wrote:
> > On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> > > ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> > > 
> > > fftools/ffmpeg: convert to a threaded architecture
> > > 
> > > Change the main loop and every component (demuxers, decoders, filters,
> > > encoders, muxers) to use the previously added transcode scheduler. Every
> > > instance of every such component was already running in a separate
> > > thread, but now they can actually run in parallel.
> > > 
> > > Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> > > JEEB to be more correct and deterministic.
> > > 
> > > > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> > > ---
> > 
> > infinite loops this: (most of the time)
> > 
> > ffmpeg -i tickets///2531/interlaced_top.wmv -t 10 -bitexact -vcodec ffv1 -level 0 -y  file..avi
> > 
> > https://samples.ffmpeg.org/ffmpeg-bugs/trac/ticket2531/interlaced_top.wmv
> 
> also a related question
> is there a way to make things execute serially in a predictive manner ?

You should be able to make it effectively serial with something like
schedtool -a 0x1 -F -p 1 -e ./ffmpeg ...
It still won't be deterministic across runs though.

-- 
Anton Khirnov
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Thilo Borgmann via ffmpeg-devel]

Am 14.12.23 um 00:16 schrieb Michael Niedermayer:
> On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
>> ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
>>
>> fftools/ffmpeg: convert to a threaded architecture
>>
>> Change the main loop and every component (demuxers, decoders, filters,
>> encoders, muxers) to use the previously added transcode scheduler. Every
>> instance of every such component was already running in a separate
>> thread, but now they can actually run in parallel.
>>
>> Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
>> JEEB to be more correct and deterministic.
>>
>>> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
>> ---

Using --assert-level=2 breaks "everything" pretty early.
I'm on 5256b2f:

thilo@MAC-2 FFmpeg % lldb -- ./ffmpeg_g -lavfi testsrc2 -f null -
(lldb) target create "./ffmpeg_g"
Current executable set to '/Users/thilo/FFmpeg/ffmpeg_g' (arm64).
(lldb) settings set -- target.run-args  "-lavfi" "testsrc2" "-f" "null" "-"
(lldb) r
Process 24543 launched: '/Users/thilo/FFmpeg/ffmpeg_g' (arm64)
ffmpeg version N-112972-g5256b2fbe6 Copyright (c) 2000-2023 the FFmpeg developers
   built with Apple clang version 12.0.5 (clang-1205.0.22.9)
   configuration: --assert-level=2
   libavutil      58. 33.100 / 58. 33.100
   libavcodec     60. 35.100 / 60. 35.100
   libavformat    60. 18.100 / 60. 18.100
   libavdevice    60.  4.100 / 60.  4.100
   libavfilter     9. 14.100 /  9. 14.100
   libswscale      7.  6.100 /  7.  6.100
   libswresample   4. 13.100 /  4. 13.100
Stream mapping:
   testsrc2:default -> Stream #0:0 (wrapped_avframe)
pthread_mutex_lock failed with error: Invalid argument
Process 24543 stopped
* thread #1, queue = 'com.apple.main-thread', stop reason = signal SIGABRT
     frame #0: 0x000000018cb3ed78 libsystem_kernel.dylib`__pthread_kill + 8
libsystem_kernel.dylib`__pthread_kill:
->  0x18cb3ed78 <+8>:  b.lo   0x18cb3ed98               ; <+40>
     0x18cb3ed7c <+12>: pacibsp
     0x18cb3ed80 <+16>: stp    x29, x30, [sp, #-0x10]!
     0x18cb3ed84 <+20>: mov    x29, sp
Target 0: (ffmpeg_g) stopped.
(lldb) bt
* thread #1, queue = 'com.apple.main-thread', stop reason = signal SIGABRT
   * frame #0: 0x000000018cb3ed78 libsystem_kernel.dylib`__pthread_kill + 8
     frame #1: 0x000000018cb73ee0 libsystem_pthread.dylib`pthread_kill + 288
     frame #2: 0x000000018caae340 libsystem_c.dylib`abort + 168
     frame #3: 0x0000000100c5d028 ffmpeg_g`sch_start.cold.2 at ffmpeg_sched.c:91:5 [opt]
     frame #4: 0x00000001000204ac ffmpeg_g`sch_start at thread.h:91:5 [opt]
     frame #5: 0x00000001000204a0 ffmpeg_g`sch_start(sch=<unavailable>) at ffmpeg_sched.c:1415 [opt]
     frame #6: 0x000000010002b2b4 ffmpeg_g`main [inlined] transcode(sch=<unavailable>) at ffmpeg.c:922:11 [opt]
     frame #7: 0x000000010002aeb8 ffmpeg_g`main(argc=<unavailable>, argv=<unavailable>) at ffmpeg.c:1050 [opt]
     frame #8: 0x0000000102c1908c dyld`start + 520

-Thilo
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Anton Khirnov]

Quoting Thilo Borgmann via ffmpeg-devel (2023-12-14 09:18:30)
> Am 14.12.23 um 00:16 schrieb Michael Niedermayer:
> > On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> >> ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> >>
> >> fftools/ffmpeg: convert to a threaded architecture
> >>
> >> Change the main loop and every component (demuxers, decoders, filters,
> >> encoders, muxers) to use the previously added transcode scheduler. Every
> >> instance of every such component was already running in a separate
> >> thread, but now they can actually run in parallel.
> >>
> >> Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> >> JEEB to be more correct and deterministic.
> >>
> >>> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> >> ---
> 
> Using --assert-level=2 breaks "everything" pretty early.
> I'm on 5256b2f:

should be fixed by the patch I just sent

-- 
Anton Khirnov
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Gyan Doshi]

On 2023-12-14 02:48 pm, Anton Khirnov wrote:
>>> On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
>>>> ffmpeg | branch: master | Anton Khirnov<anton@khirnov.net>  | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
>>>>
>>>> fftools/ffmpeg: convert to a threaded architecture
>>>>
>>>> Change the main loop and every component (demuxers, decoders, filters,
>>>> encoders, muxers) to use the previously added transcode scheduler. Every
>>>> instance of every such component was already running in a separate
>>>> thread, but now they can actually run in parallel.
>>>>
>>>> Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
>>>> JEEB to be more correct and deterministic.
>>>>
>>>>> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
>>>> ---
> should be fixed by the patch I just sent
>

I tested a single input --> output conversion to compare performance and 
I noticed the final report now displays N/A for time, bitrate and speed i.e.

frame= 1500 fps=109 q=-1.0 Lsize=   17363kB time=N/A bitrate=N/A speed=N/A

whereas commit 1439784ff0 shows

frame= 1500 fps=108 q=-1.0 Lsize=   17363kB time=00:00:59.97 
bitrate=2371.6kbits/s speed=4.34x

Is this expected?

I do see values *while* transcoding.

Regards,
Gyan
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Anton Khirnov]

Quoting Gyan Doshi (2023-12-14 12:22:51)
> 
> 
> On 2023-12-14 02:48 pm, Anton Khirnov wrote:
> >>> On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> >>>> ffmpeg | branch: master | Anton Khirnov<anton@khirnov.net>  | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> >>>>
> >>>> fftools/ffmpeg: convert to a threaded architecture
> >>>>
> >>>> Change the main loop and every component (demuxers, decoders, filters,
> >>>> encoders, muxers) to use the previously added transcode scheduler. Every
> >>>> instance of every such component was already running in a separate
> >>>> thread, but now they can actually run in parallel.
> >>>>
> >>>> Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> >>>> JEEB to be more correct and deterministic.
> >>>>
> >>>>> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> >>>> ---
> > should be fixed by the patch I just sent
> >
> 
> I tested a single input --> output conversion to compare performance and 
> I noticed the final report now displays N/A for time, bitrate and speed i.e.
> 
> frame= 1500 fps=109 q=-1.0 Lsize=   17363kB time=N/A bitrate=N/A speed=N/A
> 
> whereas commit 1439784ff0 shows
> 
> frame= 1500 fps=108 q=-1.0 Lsize=   17363kB time=00:00:59.97 
> bitrate=2371.6kbits/s speed=4.34x
> 
> Is this expected?

No, but check whether
http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2023-December/318326.html
(also depends on the previous patch) fixes it.

-- 
Anton Khirnov
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Gyan Doshi]

On 2023-12-14 04:56 pm, Anton Khirnov wrote:
> Quoting Gyan Doshi (2023-12-14 12:22:51)
>>
>> On 2023-12-14 02:48 pm, Anton Khirnov wrote:
>>>>> On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
>>>>>> ffmpeg | branch: master | Anton Khirnov<anton@khirnov.net>  | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
>>>>>>
>>>>>> fftools/ffmpeg: convert to a threaded architecture
>>>>>>
>>>>>> Change the main loop and every component (demuxers, decoders, filters,
>>>>>> encoders, muxers) to use the previously added transcode scheduler. Every
>>>>>> instance of every such component was already running in a separate
>>>>>> thread, but now they can actually run in parallel.
>>>>>>
>>>>>> Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
>>>>>> JEEB to be more correct and deterministic.
>>>>>>
>>>>>>> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
>>>>>> ---
>>> should be fixed by the patch I just sent
>>>
>> I tested a single input --> output conversion to compare performance and
>> I noticed the final report now displays N/A for time, bitrate and speed i.e.
>>
>> frame= 1500 fps=109 q=-1.0 Lsize=   17363kB time=N/A bitrate=N/A speed=N/A
>>
>> whereas commit 1439784ff0 shows
>>
>> frame= 1500 fps=108 q=-1.0 Lsize=   17363kB time=00:00:59.97
>> bitrate=2371.6kbits/s speed=4.34x
>>
>> Is this expected?
> No, but check whether
> http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2023-December/318326.html
> (also depends on the previous patch) fixes it.
>

Yes, it does.

Regards,
Gyan


_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Michael Niedermayer]

[-- Attachment #1.1: Type: text/plain, Size: 1585 bytes --]

On Thu, Dec 14, 2023 at 04:25:45AM +0100, Anton Khirnov wrote:
> Quoting Michael Niedermayer (2023-12-14 00:16:14)
> > On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> > > ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> > > 
> > > fftools/ffmpeg: convert to a threaded architecture
> > > 
> > > Change the main loop and every component (demuxers, decoders, filters,
> > > encoders, muxers) to use the previously added transcode scheduler. Every
> > > instance of every such component was already running in a separate
> > > thread, but now they can actually run in parallel.
> > > 
> > > Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> > > JEEB to be more correct and deterministic.
> > > 
> > > > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> > > ---
> > 
> > infinite loops this: (most of the time)
> > 
> > ffmpeg -i tickets///2531/interlaced_top.wmv -t 10 -bitexact -vcodec ffv1 -level 0 -y  file..avi
> > 
> > https://samples.ffmpeg.org/ffmpeg-bugs/trac/ticket2531/interlaced_top.wmv
> 
> Did you test this and the other loop with 7cc4b306eb? I cannot reproduce
> it here.

I cannot reproduce at with current git master either so maybe its fixed

thx

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

I am the wisest man alive, for I know one thing, and that is that I know
nothing. -- Socrates

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Michael Niedermayer]

[-- Attachment #1.1: Type: text/plain, Size: 2084 bytes --]

On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> 
> fftools/ffmpeg: convert to a threaded architecture
> 
> Change the main loop and every component (demuxers, decoders, filters,
> encoders, muxers) to use the previously added transcode scheduler. Every
> instance of every such component was already running in a separate
> thread, but now they can actually run in parallel.
> 
> Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> JEEB to be more correct and deterministic.
> 
> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> ---
> 
>  Changelog                                        |   2 +
>  fftools/ffmpeg.c                                 | 374 +-----------
>  fftools/ffmpeg.h                                 |  97 +--
>  fftools/ffmpeg_dec.c                             | 313 +++-------
>  fftools/ffmpeg_demux.c                           | 268 ++++-----
>  fftools/ffmpeg_enc.c                             | 377 ++----------
>  fftools/ffmpeg_filter.c                          | 714 ++++++-----------------
>  fftools/ffmpeg_mux.c                             | 324 ++--------
>  fftools/ffmpeg_mux.h                             |  24 +-
>  fftools/ffmpeg_mux_init.c                        |  88 +--
>  fftools/ffmpeg_opt.c                             |   6 +-
>  tests/ref/fate/ffmpeg-fix_sub_duration_heartbeat |  34 +-
>  12 files changed, 600 insertions(+), 2021 deletions(-)

breaks (vissible artifacts)
./ffmpeg -bitexact -t 1 -i '/home/michael/videos/bink/bikb/CGOOD3.BIK'  -y test.avi && ./ffplay test.avi

should be in game-formats/bink/bikb/

should i open a ticket ?

thx

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

I do not agree with what you have to say, but I'll defend to the death your
right to say it. -- Voltaire

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

Re: [FFmpeg-devel] [FFmpeg-cvslog] fftools/ffmpeg: convert to a threaded architecture

[Michael Niedermayer]

[-- Attachment #1.1: Type: text/plain, Size: 7574 bytes --]

On Tue, Dec 12, 2023 at 07:41:18AM +0000, Anton Khirnov wrote:
> ffmpeg | branch: master | Anton Khirnov <anton@khirnov.net> | Tue Jul 18 16:37:52 2023 +0200| [d119ae2fd82a494d9430ff4d4fc262961a68c598] | committer: Anton Khirnov
> 
> fftools/ffmpeg: convert to a threaded architecture
> 
> Change the main loop and every component (demuxers, decoders, filters,
> encoders, muxers) to use the previously added transcode scheduler. Every
> instance of every such component was already running in a separate
> thread, but now they can actually run in parallel.
> 
> Changes the results of ffmpeg-fix_sub_duration_heartbeat - tested by
> JEEB to be more correct and deterministic.
> 
> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=d119ae2fd82a494d9430ff4d4fc262961a68c598
> ---
> 
>  Changelog                                        |   2 +
>  fftools/ffmpeg.c                                 | 374 +-----------
>  fftools/ffmpeg.h                                 |  97 +--
>  fftools/ffmpeg_dec.c                             | 313 +++-------
>  fftools/ffmpeg_demux.c                           | 268 ++++-----
>  fftools/ffmpeg_enc.c                             | 377 ++----------
>  fftools/ffmpeg_filter.c                          | 714 ++++++-----------------
>  fftools/ffmpeg_mux.c                             | 324 ++--------
>  fftools/ffmpeg_mux.h                             |  24 +-
>  fftools/ffmpeg_mux_init.c                        |  88 +--
>  fftools/ffmpeg_opt.c                             |   6 +-
>  tests/ref/fate/ffmpeg-fix_sub_duration_heartbeat |  34 +-
>  12 files changed, 600 insertions(+), 2021 deletions(-)

breaks: (races, violates bitexact)

for i in `seq 50` ; do ./ffmpeg -v 0 -i ~/tickets/4003/mp3_demuxer_EOI.mp3 -bitexact -vframes 1 -y /tmp/file-4003.avi ; md5sum /tmp/file-4003.avi

before:
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi


aftre:
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
db7ad90e0e02af90e98e09c9b0158124  /tmp/file-4003.avi
933c2cc12cda1deac7245ec768f5b35f  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi
1508e16c1877abe172173cd1e11f1155  /tmp/file-4003.avi


should i open a ticket to track this or someone has time
to look into this ?

thx

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

I do not agree with what you have to say, but I'll defend to the death your
right to say it. -- Voltaire

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".