On 13 May 2014 23:34, Luca Barbato <[email protected]> wrote:

> On 13/05/14 17:25, [email protected] wrote:
> > From: James Yu <[email protected]>
> >
> > Using NEON intrinsics to support ARMv7 and ARMv8 simultaneously.
> > Rewrite ARMv7 assembly to intrinsics version.
> >
> > Toolchain:
> > armv7: gcc-linaro-arm-linux-gnueabihf-4.8-2014.01_linux
> > armv8: gcc-linaro-aarch64-linux-gnu-4.8-2013.10_linux
> >
> > [RFC] Currently, libavcodec had a aarch64 dir includes all related
> assembly
> > code. However, I put this file under libavcodec/arm dir first, since this
> > support ARMv7 and ARMv8 simultaneously. And RFC how could we port and
> > optimize libav to aarch64.
>
> Is the speed difference negligible ? I recall that for arm gcc (and
> clang) weren't exactly great at mapping intrinsics and that produced
> less than optimal code?
>
> Yes, it sometimes have different performance than assembly code.

I try benchmark by using Linaro toolchain for this patch, And run below
commend
"avconv -i mp4_movie -an -f null EMPTY-OUTPUT"
to get the fps. it doesn't have obvious different performance.


> Surely it is a step in the right direction (C -> intrinsics -> faster asm).
>
> I wish we have smarter compilers =)
>
> Thanks a lot for the patch, I let Janne review it.
>
> Thanks :-)

--
Regards,
James


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

Reply via email to