[FFmpeg-devel] [PATCH 09/13] avcodec/svq1dec: clear MMX state after MB decode loop

[Hendrik Leppkes]

On Wed, Oct 26, 2016 at 4:52 PM,  <u-9iep at aetey.se> wrote:
> On Wed, Oct 26, 2016 at 04:21:14PM +0200, Hendrik Leppkes wrote:
>> You can add "not caring about first-gen sse2 CPUs" to the list as
>> well, if you want. Those are way old as well.
>> There is going to be a performance loss either way, except that emms
>> slows it down everywhere, while using sse2 is likely to be fine on
>> modern CPUs. So IMHO thats the better course to take, if any.
>
> Note that making MMX-acceleration unavailable would hurt at most on
> those slowest, non-sse systems, which otherwise e.g. often struggle for
> real-time decoding of heavier streams.
>
> One of the highly appreciated virtues of ffmpeg is its efficiency,
> this makes hardware much more useful. Please do not cut off the low
> end systems when possible.
>

Its not about low-end, its about 15+ years old hardware. At some point
you just have to replace this stuff.
For $100 you can buy a system at least 10 times as fast then those.
Heck even a RPi might rival those for $20 or so.

If you want to convince developers to fix the MMX situation properly,
which would cost a lot of time to do, then the best course is to look
forward, which the SSE2 option imho is doing.
All other options are just band-aids.

- Hendrik