Message-ID: <33203E53.22DA@eik.bme.hu>
Date: Fri, 07 Mar 1997 17:12:03 +0100
From: "DR. Andras Solyom" <solyom AT DELeik DOT bme DOT hu>
Reply-To: solyom AT eik DOT bme DOT hu
MIME-Version: 1.0
To: djgpp <djgpp AT delorie DOT com>
Subject: Interesting benchmark results
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

I downloaded a small benchmark program named 'Java Fhourstones 2.0
connect-4 solver' from 
	http://www.cwi.nl/~tromp/c4/fhour.html
because I was interested in a speed comparison between my PC and our
Power PC processor based IBM RISC 6000 model 250 workstation.
This benchmark is claimed to be memory intensive. My results were: 

System                        rel. speed (higher numbers are better)
....................................................................
Pentium 100, Win95, DJGPP2, -O         46.3
                            -O3        45   (slower ?? )
             DOS6.22        -O         49
                            -O3        49.9
RISC 6000, AIX, XLC         -O         48


I found it very interesting that the program run at almost the same 
speed in a DOS window under Win95, than in clean DOS. And it is very
impressive that my PC with DJGPP can be as fast or faster than our
workstations... (I could not check this program with my Borland
compiler, because it does not have the 64msec resolution timer and 64
bit integers that this program needs.) 
Can anyone test other compilers so that we can compare them with DJGPP?
I can mail the 4Kbytes gzip-ed tar file to anyone interested, who cannot
get it with http on Monday)


					Andras