Re: geomview performance

[slevy]

> From csilva at cs.sunysb.edu Thu May 12 15:55:08 1994
> Date: Thu, 12 May 1994 16:54:28 -0400
> From: csilva at cs.sunysb.edu (Claudio Silva)
> To: slevy at geom
> Subject:  geomview performance
> 
> A while back, you sent performance numbers for geomview under different 
> machine configurations. I wonder how you got the numbers ? and also
> do you know of the performance improvement of the Pentium over
> the 486 ? It should be pretty high as the Pentium has a pretty fast
> FPU.   
> 
> Thank you in advance,
> 
> Claudio.

I made up three tests: repeatedly drawing one of several things, running
geomview with a fixed window size, etc., and using geomview's internal
clock to report timing.

You can pick up priv/slevy/geomview-speedtest.tar.Z from geom.umn.edu
to get the tests and the results gathered so far.
to get the tests and the results gathered so far.  I've gotten one report
from someone with a Pentium machine; here it is compared to our '486 machine:

# Tests: "polygon" : 1620 small flat-shaded LIGHTED polygons
# 	 "matrix"  : 30 vectors x 115 matrices
#	 "tmesh"   : flat-shaded LIGHTED tmesh sphere: 35 rows of 70 tri's
#			(Uses tmesh primitive on SGI; other platforms just
#			 draw 35x35 quadrilateral mesh)
# All tests in 200x200 window; all polygons are flat-shaded and LIGHTED.
# Times are milliseconds per test cycle.

	       polygon	matrix tmesh
Gateway '486 PC  425     1600   250     kowalevsky (486 66MHz PC, NS 3.1)
ALR Pentium PC   230      912   145     60MHz Pentium, NS 3.1, ALR Evolution V