On Fri, 11 Jul 2014 21:48:36 +0100, Diego Biurrun <[email protected]> wrote:
You're optimizing with ARM in mind but testing on x86?

As has been said, yes it's a pure C optimisation, so checking for
validity can generally be done on any platform. In my experience, FATE is
much easier to build and run with a native compiler, but compiling libav
natively on the ARM I'm targeting takes several hours so I try to avoid
doing so!

The reason why I say it's with special attention to ARM is because of the
typical size of a frame (and therefore the distance between start codes)
in an HD VC-1 stream, and how that relates to the cache sizes on the ARM
(it's bigger than even the L2 cache). This explains why it's such a
performance drain to perform multiple passes across the data between
neighbouring start codes.

I expect there is still a performance gain on other architectures, just
not necessarily as extreme. The benchmark figures I gave were for ARM, if
it wasn't clear.

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

Reply via email to