Hi,

On Fri, Jul 27, 2012 at 5:48 PM, Ronald S. Bultje <[email protected]> wrote:
> On Fri, Jul 27, 2012 at 2:43 PM, Måns Rullgård <[email protected]> wrote:
>> However, the question still remains why it is in generic code.
>
> That's hard to say in hindsight, but it seems it was for simplicity so
> that you don't have to add it to each individual mmx function, thus
> making the asumption they would all use movntq. See also (directly
> under the #endif, just outside the context in this patch) its use of
> EMMS, even if it only called SSE2 functions and thus the MMX state was
> never clobbered...

Ping on the patch itself also - can we keep it, at least for now? I
don't midn refactoring sfence out at some point, but that's a separate
task from the point of this patch.

Ronald
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to