[FFmpeg-devel] looking for comparison of intrinsics vs hand written asm

[Loren Merritt]

On Fri, 30 Oct 2009, Frank Barchard wrote:

> asm has 2 advantages
> 1. more control over order, letting you tune to architecture.  ie you can
> pair instructions.

You don't need to tune to a specific cpu in order to beat the compiler's 
scheduling. Just write the asm in any old naive order, and it will still 
be simultaneously faster on every cpu than auto-scheduled intrinsics.
(As long as the cpu does at least a little out-of-order execution. Otoh, 
if it's strictly in-order, then maybe the compiler can do better than no 
scheduling at all; I've never tried to optimize for such a case.)

--Loren Merritt