Path: utzoo!attcan!uunet!lll-winken!ames!mailrus!iuvax!rutgers!uwvax!per2!dag
From: dag@per2.UUCP (Daniel A. Glasser)
Newsgroups: comp.sys.atari.st
Subject: Re: Turbo C
Summary: MWC Dhrystones not that bad
Keywords: Calamus, Turbo C
Message-ID: <815@per2.UUCP>
Date: 10 Jan 89 15:03:40 GMT
References: <834@sdcc15.ucsd.edu> <537@stag.math.lsa.umich.edu> <3972@mtuxo.att.com>
Organization: Persoft Inc., Madison, WI
Lines: 36

In article <3972@mtuxo.att.com>, rcd@mtuxo.att.com (XG1V6-R.DUTT) writes:
> On a MEGA 2 ST (68000 CPU and 8 MHz clock) Turbo C (1.0) achieved a throughput
> of 1730 Dhrystones/sec. By comparison, Aztec C (3.6A) clocks in at 
> 851 Dhry/sec on the same system, which is probably typical of most other 
> compilers. 
> 
> -Rana Dutt

Actually, the Mark Williams C compiler for the Atari ST have never been down
in the 851 dhrystones/sec range --

	MWC Version:	non-register	register
	   1.1		   914		  946
	  2.1.5		  1109		 1172
	   3.0		  1270		 1343

Results better than 1400 d/s were obtained with special tweeking of a few
switches, but that is cheating.

The only numbers I have for MegaMax and Lattice are from old versions

	Compiler	non-register	register
	MegaMax C	    826		  862
	MegaMax C w/opt	    834		  870
	MCC Lattice C	    524		  530	first release MCC Lattice C
	MCC Lattice C	    677		  676	shorts used in place of ints

Maybe someone could post more recent numbers for Laser (formerly MegaMax),
Lattice, Gnu, Sozobon, and any others I've forgotten (or don't know about).
If you need a copy of the Dhryston benchmark code send me mail and I'll see
what I can do.  Use the address below as the per2 address may go away soon.
-- 
 _____________________________________________________________________________
    Daniel A. Glasser                           One of those things that goes
    ...!persoft!dag                             "BUMP!!!(ouch)" in the night. 
 ---Persoft, Inc.---------465 Science Drive-------Madison, WI 53711-----------