On Fri, Jul 21, 2017 at 12:18:53PM +0200, Tobias Rapp wrote: > On 21.07.2017 02:47, Michael Niedermayer wrote: > >On Fri, Jul 21, 2017 at 01:04:03AM +0200, Nicolas George wrote: > >>Le duodi 2 thermidor, an CCXXV, Michael Niedermayer a écrit : > >>>breaks on mips (probably big endian issue) > >> > >>Breaks on MIPS, or detects that something is already broken in MIPS and > >>nobody noticed until now? > > > >this patch breaks the selftests on mips or maybe more precissely > >it adds a test which fails > > > > > >> > >>Could you perhaps re-run the test with V=1, then re-run the command line > >>with -f png, to see how the output is broken? > > > >Stream #0:0: Video: rawvideo ([12][0]3Y / 0x5933000C), yuv444p12be, 320x240 > >[SAR 1:1 DAR 4:3], q=2-31, 13824 kb/s, 5 fps, 5 tbn, 5 tbc (default) > > > >the format used seems the native one that has different endianness > >and thus different checksums > > > >md5sum out.png mips/out.png > >012e1906fe84ee6f757ef3f1473cc26d out.png > >012e1906fe84ee6f757ef3f1473cc26d mips/out.png > > > >i tried forcing le but it alone didnt give the same result > >ive to sleep now so i cant look further into this today > >but iam happy to test patches tomorrow, also iam just testing with > >mips qemu not real hw so if it wasnt for the one time work to set > >build env up and qemu than it would be rather easy for anyone to test > > I tested the opposite direction by forcing yuv444p12be on x86-64 and > got different frame-CRCs "0xda6a937e" compared to the MIPS-CRCs you > posted earlier "0xba079be7". Not sure what is happening ...
mybe the le<->be is the BE output you get identical to LE ? (besides the endianness) [...] -- 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
signature.asc
Description: Digital signature
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
