[FFmpeg-devel] [FFMpeg-Devel][GSoC][PATCH 1/2] postproc: Updated postprocess_template to use new sse/avx deinterlace functions

[Michael Niedermayer]

On Thu, Apr 23, 2015 at 12:20:37AM -0400, Tucker DiNapoli wrote:
> A few notes on changes from the last version of this patch.
> The main issue with the previous code was with the sse2/avx2
> implementation of the blockCopy function, so for the time being the MMX2 version
> is used instead. I tried to place the MMX2 version into a function, but this 
> did not work, my best guess as to why is alignment, but I really don't know. 
> The way it's done now is a bit ugly but it works and I don't have time to figure
> out the issue right now.
> 
> This commit adds several new files containing yasm assembly code, they are:
> PPContext.asm; Defines the PPContext struct using the yasm struc command
> PPUtil.asm; Various utility macros used in the other asm code
> block_copy.asm; Implements the block copy function, the sse2 and avx2
> versions copy multiple blocks at once.
> deinterlace.asm; Contains implemenations of the postprocessing filters
> with support for sse2 and avx2.
> 
> Adding these new functions to postprocess_template entailed adding a new
> templates for AVX2 and modifying the current SSE2 template to use the
> sse2 functions.  A new deinterlace function was added to move the logic
> of which deinterlace function to use out of the postprocess function and
> make adding the new functions eaiser. The inline code for packing QP
> into pQPb was moved into a seperate asm file and uptaded for sse2/avx2.
> 
> Currently the sse2/avx2 deinterlace filters don't give results which are
> bitexact to the C results, so I've modified one of the postprocessing
> tests so that only the C funcitons are tested by fate. Ultimately either
> the sse2/avx2 code will need to be fixed or diffrent tests will need to
> be added. I'm not sure if this is a problem with my code, a problem inherent
> in using sse2/avx2 or a problem that's a result of deinterlacing being done
> blockwise.

On a AVX machine (using SSE2 i assume)
-vf pp=md
-vf pp=lb
-vf pp=ci
-vf pp=fd
-vf pp=l5

tested with ./ffplay matrixbench_mpeg2.mpg -vf tinterlace=4,pp=XX

shows blocking artifacts as if some deinterlace function is applied
to only every second column of some size
thats not just not bitexact that is completely not working

On a AVX2 machine the tests above produce an output that
looks as if one out of 4 columns is filtered with  the other 3 still
showing interlacing artifacts


also the code produces these warnings
libpostproc/x86/PPContext.asm:37: warning: section flags ignored on section redeclaration
libpostproc/x86/PPContext.asm:77: warning: section flags ignored on section redeclaration
libpostproc/x86/PPUtil.asm:215: warning: section flags ignored on section redeclaration

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

Old school: Use the lowest level language in which you can solve the problem
            conveniently.
New school: Use the highest level language in which the latest supercomputer
            can solve the problem without the user falling asleep waiting.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: Digital signature
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20150423/77b3a70e/attachment.asc>