From: broehl@sunee.uwaterloo.ca (Bernie Roehl)
Subject: Re: TECH: REND386 and the 486DX
Date: Tue, 15 Dec 1992 12:14:48 GMT
Message-ID: <BzAvCp.4vF@watserv1.uwaterloo.ca>
Organization: University of Waterloo


>First of all, REND386 only runs ALMOST twice
>as fast on my 486/50DX as it does on my 486/25SX.

Right, that makes sense.  At twice the clock speed, it should be almost twice
the performance.

There will also be differences in speed attributable to different video
cards; in some (many) cases, the bottleneck is in writing to video ram.

>It's obviously not
>using the 50DX to it's fullest potential (on the benchmark program 'Checkit'
>the 25SX scores 226K drystones (math speed), the 50DX scores 8200K drystones

Those are probably done using the floating-point capabilities of the 486.
REND386 uses integer math only (which will be faster than floating point,
even on a 486).  It therefore doesn't require or use a floating point math
unit.

>How can I get it to use the coprocessor's
>instruction set?  Can I recompile it differently to get it to do so?

You could rewrite the code to use floating point, but it would be a lot of
work and the result would actually run slower.

>And second...does anybody know ANYTHING about the VAX system (I know
>NOTHING about Vax...ABSOLUTELY NOTHING) ...I have an opportunity to
>get one that's about 6 years old...can I use it for anything? (i.e.
>networked with my 486's to take care of some aspect of VR?)

Hmm.  I suspect that VAXen are no longer worth bothering with, given the
speed of current systems.  Six years is a really long time in this industry.

Also bear in mind the cost of additional memory for those systems (assuming
it's still available!), the cost of peripherals, software, etc etc...

I wouldn't bother with it.

-- 
	Bernie Roehl, University of Waterloo Electrical Engineering Dept
	Mail: broehl@sunee.UWaterloo.ca
	BangPath: uunet!watmath!sunee!broehl
	Voice:  (519) 885-1211 x 2607 [work]