New submission from Vitor <vitor1...@gmail.com>:
For example http://samples.mplayerhq.hu/fate-suite/4xm/version1.4xm . This is
blocking fate testing for this format.
Note that the RGB565 -> RGB24 conversion is not the culprit, I've tested using
RGB565NE as output and bitswapping for comparing and they still differs.
In x86 (Atom)
vi...@vitor-laptop:~/ffmpeg/ffmpeg$ ./ffmpeg -i ../fate-suite/4xm/version1.4xm
-an -pix_fmt rgb24 -f md5 -
FFmpeg version SVN-r24953, Copyright (c) 2000-2010 the FFmpeg developers
built on Aug 28 2010 13:05:28 with gcc 4.4.3
configuration: --cc='ccache gcc' --cpu=host
libavutil 50.24. 0 / 50.24. 0
libavcore 0. 6. 0 / 0. 6. 0
libavcodec 52.87. 0 / 52.87. 0
libavformat 52.78. 3 / 52.78. 3
libavdevice 52. 2. 1 / 52. 2. 1
libavfilter 1.38. 1 / 1.38. 1
libswscale 0.11. 0 / 0.11. 0
[4xm @ 0x9200470] Estimating duration from bitrate, this may be inaccurate
Input #0, 4xm, from '../fate-suite/4xm/version1.4xm':
Duration: 00:00:04.87, start: 0.000000, bitrate: 705 kb/s
Stream #0.0: Video: 4xm, rgb565le, 640x480, 15 tbr, 15 tbn, 15 tbc
Stream #0.1: Audio: pcm_s16le, 22050 Hz, 2 channels, s16, 705 kb/s
[buffer @ 0x92040e0] w:640 h:480 pixfmt:rgb565le
[ffmpeg_output @ 0x9204320] auto-inserting filter 'auto-inserted scaler 0'
between the filter 'src' and the filter 'out'
[scale @ 0x9205e00] w:640 h:480 fmt:rgb565le -> w:640 h:480 fmt:rgb24
flags:0xa0000004
Output #0, md5, to 'pipe:':
Metadata:
encoder : Lavf52.78.3
Stream #0.0: Video: rawvideo, rgb24, 640x480, q=2-31, 200 kb/s, 90k tbn, 15
tbc
Stream mapping:
Stream #0.0 -> #0.0
Press [q] to stop encoding
MD5=404d25e878508c6da3ee1975d21b732b
frame= 15 fps= 0 q=0.0 Lsize= 0kB time=1.00 bitrate= 0.3kbits/s
video:13500kB audio:0kB global headers:0kB muxing overhead -99.999732%
In PPC
vi...@saracen ~/ffmpeg $ ./ffmpeg -i
/misc/samples/mphq/fate-suite/4xm/version1.4xm -an -pix_fmt rgb24 -f md5 -
FFmpeg version SVN-r24220, Copyright (c) 2000-2010 the FFmpeg developers
built on Jul 12 2010 21:42:07 with gcc 4.3.5
configuration: --cc='ccache gcc'
libavutil 50.22. 0 / 50.22. 0
libavcodec 52.83. 0 / 52.83. 0
libavformat 52.73. 0 / 52.73. 0
libavdevice 52. 2. 0 / 52. 2. 0
libavfilter 1.22. 0 / 1.22. 0
libswscale 0.11. 0 / 0.11. 0
[4xm @ 0x10a79d20] Estimating duration from bitrate, this may be inaccurate
Input #0, 4xm, from '/misc/samples/mphq/fate-suite/4xm/version1.4xm':
Duration: 00:00:04.87, start: 0.000000, bitrate: 705 kb/s
Stream #0.0: Video: 4xm, rgb565be, 640x480, 15 tbr, 15 tbn, 15 tbc
Stream #0.1: Audio: pcm_s16le, 22050 Hz, 2 channels, s16, 705 kb/s
[ffmpeg_output @ 0x10a7dc10] auto-inserting filter 'auto-inserted scaler 0'
between the filter 'src' and the filter 'out'
[scale @ 0x10a7f660] w:640 h:480 fmt:rgb565be -> w:640 h:480 fmt:rgb24
flags:0x10000004
Output #0, md5, to 'pipe:':
Metadata:
encoder : Lavf52.73.0
Stream #0.0: Video: rawvideo, rgb24, 640x480, q=2-31, 200 kb/s, 90k tbn, 15
tbc
Stream mapping:
Stream #0.0 -> #0.0
Press [q] to stop encoding
MD5=e3669ab08d8c7341ad1dfb93c493426d 0kB time=0.87 bitrate= 0.0kbits/s
frame= 15 fps= 0 q=0.0 Lsize= 0kB time=1.00 bitrate= 0.3kbits/s
video:13500kB audio:0kB global headers:0kB muxing overhead -99.999732%
----------
messages: 11774
priority: normal
status: new
substatus: new
title: 4XM decoder is not bit-exact across archs for no good reason
type: bug
________________________________________________
FFmpeg issue tracker <iss...@roundup.ffmpeg.org>
<https://roundup.ffmpeg.org/issue2195>
________________________________________________
