Subject: Re: [linux-audio-dev] mutex madness
From: David Olofson (audiality_AT_swipnet.se)
Date: pe loka 01 1999 - 19:17:55 EDT
On Fri, 01 Oct 1999, Paul Barton-Davis wrote:
> BTW, several important DSP chips don't have conditional instructions
> at all. Fast ? Yes. Useful ? That's a long story ...
For those who are interested: Conditional processing can be done as in
mathematics; through expressions that "select" values depedning on other
values.
What's the point? (As it usually results in slower code...) Simpler DSP cores
and one single flow of control. The later can be very useful when you have just
about enough power to get the job done. (Which is what you'll have in a low
power system, for example.)
BTW, MMX use something in between these methods; instructions that test for
certain conditions and output masks of the full word size. The reason why that
is needed is that you cannot split the control flow when 2, 4 or 8 "channels" of
processing are using the same instructions at once...
//David
·A·U·D·I·A·L·I·T·Y· P r o f e s s i o n a l L i n u x A u d i o
- - ------------------------------------------------------------- - -
·Rock Solid David Olofson:
·Low Latency www.angelfire.com/or/audiality ·Audio Hacker
·Plug-Ins audiality_AT_swipnet.se ·Linux Advocate
·Open Source ·Singer/Composer
This archive was generated by hypermail 2b28 : pe maalis 10 2000 - 07:27:12 EST