Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!cmcl2!brl-adm!umd5!uvaarpa!virginia!krebs!wrp
From: wrp@krebs.acc.virginia.edu (Wm Pearson)
Newsgroups: comp.sys.ibm.pc
Subject: MSC5.0 benchmarks
Message-ID: <274@krebs.acc.virginia.edu>
Date: Wed, 18-Nov-87 14:26:29 EST
Article-I.D.: krebs.274
Posted: Wed Nov 18 14:26:29 1987
Date-Received: Sat, 21-Nov-87 15:11:17 EST
Organization: School of Medicine, University of Va.
Lines: 38


Newsgroups: comp.sys.ibm.pc
Subject: MSC5.0 benchmarks
Expires: 
References: 
Sender: 
Reply-To: wrp@biochsn.acc.virginia.edu (William R. Pearson)
Followup-To: 
Distribution: usa
Organization: University of Virginia, Charlottesville
Keywords: 

	I received my copy of the MSC 5.0 Update on Monday, November 16.
The packing slip indicates that it was shippe on Nov. 4.
The package comes on 9 disks; one file on one of the disks (a library)
was unreadable.  Here are some early benchmarks, to put the advertising
hype in perspective:

	Using the dhrystone 1.1 benchmark, Turbo'C' produced 1785
dhrystones/sec (28 sec), MSC 5.0 (CL dhry.c) 1666 and (CL -Ox -Gs dhry.c)
1724 (29 sec).  On a rapid protein database searching program which
compares a 200 residue sequence with a library containing 1.6 X10^6
residues, Turbo'C' required 4:20 while MSC required 4:05.  MSC 5.0
compiled a large package of code previously written for MSC 4.0 (and
Turbo'C') with no problems.

	So Turbo'C' still has the edge, because of compilation time
and cost.  I like to distribute source code that other people can read
and modify, with Turbo'C' they can compile it for <$70.00.  (It also
requires a lot less disk space). But MSC
does contain a graphics library.  Unfortunately the MSC graphics library
appears to not support the hercules graphics cards.

	I wonder what Turbo'C' 2.0 will look like.

Bill Pearson
wrp@virginia.EDU
...!uunet!virginia, as