Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!ucsd!ucbvax!galileo.berkeley.edu!jbuck
From: jbuck@galileo.berkeley.edu (Joe Buck)
Newsgroups: comp.benchmarks
Subject: Re: c.b.
Message-ID: <39473@ucbvax.BERKELEY.EDU>
Date: 9 Nov 90 01:03:08 GMT
References: <7571@eos.arc.nasa.gov> <7575@eos.arc.nasa.gov>
Sender: usenet@ucbvax.BERKELEY.EDU
Reply-To: jbuck@galileo.berkeley.edu (Joe Buck)
Lines: 16

In article <7575@eos.arc.nasa.gov>, eugene@eos.arc.nasa.gov (Eugene Miya) writes:
|> We have far too few of the former in this country.  Steve Lundstrom
|> pointed out that only those with supercomputers are really concern with
|> cycle counting (generalized of course).

Not so, Eugene.  Those of us who do real-time DSP (for example, I used
to do low-bit-rate speech for a living) count every cycle and get very
upset at pipeline bubbles; we also have tiny memories to work with in
many cases so we count every word.

If anyone wants to discuss this, take it to comp.arch or comp.dsp or
something.

--
Joe Buck
jbuck@galileo.berkeley.edu	 {uunet,ucbvax}!galileo.berkeley.edu!jbuck