[x264-devel] Multislicing support

[Etienne Bömcke]

Thanks for your comments.

I corrected the variable names and converted the tabs, now I'm ready  
to run a few tests with the patch as you recommended. However, I can't  
compile x264 anymore with Xcode under Leopard. Logs report a problem I  
read about in this mailing list : the file cabac-a.asm fails to compile.
I don't wan to cross-post, so I won't ask for a solution for this  
particular problem, but does anyone know which is the latest version  
to compile flawlessly under Leopard? Now I'm stuck with this version  
which doesn't compile anymore, so I can't run any test...

Etienne



On 22 Apr 2008, at 20:24, Alex Izvorski wrote:

> Hi Etienne,
>
> I've looked at it, and can confirm that at least with a couple of
> combinations of settings it produces a good stream for jm13.2 and  
> ffmpeg
> decoding.  The only changes I would suggest is some small tweaks to
> variable names for consistency (slice_size_threshold =>
> i_slice_size_threshold, b_l => b_left?), and also perhaps you can
> convert it to use spaces instead of tabs?
>
> I haven't benchmarked single-slice encoding with vs without the patch,
> nor have I tried using a mb-slicesize that would end slices in mid-row
> (a special corner case would be mb-slicesize=1).  Perhaps you or  
> someone
> could try and report your results?
>
> Aside from those issues, I would like to see it applied in git, if
> nobody objects.  Good work.
>
> Regards,
> --Alex
>
>
> On Tue, 2008-04-22 at 12:18 +0200, Etienne Bömcke wrote:
>> Hi guys,
>>
>> I'd like to get your attention one last time on the topic of multi-
>> sliced encoding. You'll find attached to this message the patch I
>> developped based on Mojtaba Hossein's work, adding support for
>> multiple slices during the encoding process. A new command-line  
>> option
>> has been added (--mb-slicesize) allowing one to specify the max  
>> number
>> of macroblock to be encoded in each slice, while the --no-mv-inter-
>> slices option tell x264 to output totally independent slices, both
>> spatially and temporally. Last time I sent this patch, I just  
>> received
>> a single reaction from Jason Martens, but no real comment on the  
>> patch
>> itself.
>> Could you please take a minute and look at it? I myself did some  
>> tests
>> to evaluate the performance of the patched code, but as my knowledge
>> of the encoder is limited I can't tell for sure that the patched code
>> doesn't cause significant performance losses.
>> If I don't receive any feedback, I'll assume the patch is ready to be
>> included in the official version, and I'll make sure this will be  
>> done
>> as soon as possible.
>>
>> Best regards,
>>
>> Etienne
>>
>>
>>
>>
>> --
>> Etienne Bömcke
>> UCL - Laboratoire TELE
>> etienne.bomcke at uclouvain.be
>>
>>
>>
>> _______________________________________________
>> x264-devel mailing list
>> x264-devel at videolan.org
>> http://mailman.videolan.org/listinfo/x264-devel
>
> _______________________________________________
> x264-devel mailing list
> x264-devel at videolan.org
> http://mailman.videolan.org/listinfo/x264-devel

--
Etienne Bömcke
UCL - Laboratoire TELE
etienne.bomcke at uclouvain.be



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.videolan.org/pipermail/x264-devel/attachments/20080423/4b40f90a/attachment.htm