[x264-devel] Re: OSX Intel breakage..

[Brian Raymond]

I initially started trying to fix it with this bug but ran into the  
same crash reported in this thread with the more recent x264  
versions. Since I didn't see anymore traffic on it I thought we were  
still having issues on OSX. I apparently missed the VLC patch below  
so I need to give that a shot and see what comes of it.

- Brian


On Nov 2, 2006, at 12:10 PM, Andrea Gianarro wrote:

> I had the same problem, commenting some lines as reported in this  
> thread:
> http://www.via.ecp.fr/via/ml/x264-devel/2006-06/msg00003.html
> solved everything. There is no need of special assembly for MacOS X  
> Intel.
> Regards,
>
> Andrea Gianarro
>
> On 11/2/06, Eric Petit <eric.petit at lapsus.org> wrote:
>> This patch works for me: http://trac.videolan.org/vlc/changeset/16701
>> (fixes the illegal instruction in pixel-sse2.asm due to pw_1/ssim_c1/
>> ssim_c2/mask_ff not being aligned).
>> I'm not sure why it wasn't included?
>>
>> On Nov 2, 2006, at 3:29 PM, Brian Raymond wrote:
>> > It's been a few weeks now and I've had no luck attempting to
>> > address the SSE assembly problem on Intel Macs. I was wondering who
>> > else is still working it and what progress might have been made?
>> >
>> > - Brian
>> >
>> >
>> > On Oct 3, 2006, at 2:07 PM, Andrea Gianarro wrote:
>> >
>> >> On 10/3/06, Tim Greatrex <tim.greatrex at mac.com> wrote:
>> >>> Not sure if this is right at all, as im not too hot on this kinda
>> >>> stuff and havent really looked, but do the intel macs feature the
>> >>> intel instruction set as well as the altivec for apple?  
>> (thought i
>> >>> read something about them featuring altivec somewhere)
>> >>>
>> >>> if so then it may be trying to encode with both optimisations in
>> >>> x264?
>> >>>
>> >>> Probably not tho :#)
>> >>>
>> >>> Tim
>> >>
>> >> The Intel macs feature the Intel instruction set up to SSE3.  
>> They can
>> >> run Altivec instructions in PPC binaries through the Rosetta
>> >> emulation
>> >> layer. In addition Apple recommends to use the Accelerate
>> >> Framework, a
>> >> set of libraries vector-accelerated and optimized for both  
>> Intel and
>> >> PPC (MMX, SSE, Altivec).
>> >>
>> >> Andrea Gianarro
>> >>
>> >> --
>> >> This is the x264-devel mailing-list
>> >> To unsubscribe, go to: http://developers.videolan.org/lists.html
>> >>
>> >
>> > --
>> > This is the x264-devel mailing-list
>> > To unsubscribe, go to: http://developers.videolan.org/lists.html
>> >
>>
>> --
>> This is the x264-devel mailing-list
>> To unsubscribe, go to: http://developers.videolan.org/lists.html
>>
>>
>
> -- 
> This is the x264-devel mailing-list
> To unsubscribe, go to: http://developers.videolan.org/lists.html
>

-- 
This is the x264-devel mailing-list
To unsubscribe, go to: http://developers.videolan.org/lists.html