Subject: Re: Math Libs
Date: Mon, 30 May 94 17:39:21 CDT
From: wilks@lbm.com (Stephen Wilkinson)

rudd@plk.af.mil (Douglas Rudd) said:

> I've seen a mixed bag of results on the list. Most seem to share my 
> experience; that is to say, no speed up what so ever - scanline or 
> ray trace. I've tested these libs on my GVP 040 (A2000) and GVP 030/50 
> (A1200) with much the same disapointing results. On a very long ray trace 
> (6+ hours), the stock libs are actually a couple minutes faster than the 
> 040 libs with the 882 libs falling in between.
[snip]

I would hazard a guess that Impulse compile Imagine to have the 68881/2
calls directly in the code, rather than opening the system libraries.
Your time difference (on a '30) would be mostly due to drift from the
system tasks and other system loads. The upshot of this would be that
Imagine won't be any faster with new libs. This is what I see on my
A3000 at home. Anybody got SNOOP-DOS to verify this?

Stephen
____________________________________
Stephen Wilkinson     wilks@lbm.com      "Programming is like pinball. The
Software Engineer                         reward for doing it well is the
LB&M Associates                           opportunity to do it again."
____________________________________
#define OPINION        (myown)
#define COMPANYOPINION (~myown)

      This message created using 87% recycled neuro-transmitters


                           -=> RETURN TO CONTENTS!<=-