Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!samsung!sol.ctr.columbia.edu!ira.uka.de!smurf!subnet.sub.net!mcshh!tilmann
From: tilmann@mcshh.hanse.de (Tilmann Reh)
Newsgroups: comp.os.cpm
Subject: Re: CP/M Benchmarks
Message-ID: <8487@mcshh.hanse.de>
Date: 8 Dec 90 19:00:49 GMT
Lines: 19


Doug Braun writes about CP/M Benchmarks.
He's absolutely right: the problem is that there is no standard (not only C)
compiler for CP/M. So, for testing the machine itself, we should at least
take the most usual compiler around here. I guess, that is Turbo-Pascal 3.0.
Perhaps the MBasic-Interpreter (though a hack) could serve for this purpose,
too.
So, if we use a standard set of test routines (i.e. for integer and real
arithmetic, and for I/O on ramdisk/harddisk/floppy) and compile them with
exactly the same compiler (Turbo-3), the results should be comparable.
Although I use a Z280 too, I think it's not right to use modified compilers
for benchmarks (except when the compliers are exactly the same, and we
just want to compare some Z280 with each other).
In that means, I'm using such a benchmark (published in c't magazine some
years ago) for a long time. I might post the sources on the net, if there
is interest.

Tilmann Reh			tilmann@mcshh.uucp
				tilmann@mcshh.hanse.de