Subject: Re: [linux-audio-dev] MMX, SSE, SSE2, 3DNOW
From: Steve Harris (S.W.Harris_AT_ecs.soton.ac.uk)
Date: Mon Feb 18 2002 - 21:14:28 EET
On Mon, Feb 18, 2002 at 05:34:19 +0100, Frank Neumann wrote:
> > You've all forgotten one, sane people don't had generate SIMD
> > instructions, they let a vectorising compiler do it for them, and the
> > compiler doesn't understand LADSPA, so won't know where to put the
> > conditionals.
>
> Ok, let's just assume for a second we're all not quite sane around here,
> ok? :-) I was in fact thinking of developers hand-crafting optimized
> versions of their run() or run_adding() functions, using the often-seen
> __asm__ __volatile {...} construct. I mean, there must be a reason why
MMX isn't particularly useful for c.2002 audio work.
SSE+SSE2 are significantly more complicated than MMX, there are some
people who hand craft the codes, but you have to be aware of lots of
sheduling issues, and from what I've heard modern compilers do a better
job.
Speaking for myself I have no intention of writing SSE by hand, I feel
it's a big waste of time, and penalises people with decent (ie. non x86)
CPU's. OTOH I plan to distrubte SSE accelerated binaries of my plugins
generated with the intel c compiler.
- Steve
This archive was generated by hypermail 2b28 : Mon Feb 18 2002 - 21:05:04 EET