Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site x.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!cybvax0!frog!x!john
From: john@x.UUCP (John Woods)
Newsgroups: net.arch,net.lang.c,net.micro,net.micro.pc,net.micro.68k
Subject: Re:^3 Need 286 "C" benchmark
Message-ID: <527@x.UUCP>
Date: Wed, 5-Jun-85 12:12:45 EDT
Article-I.D.: x.527
Posted: Wed Jun  5 12:12:45 1985
Date-Received: Sat, 8-Jun-85 02:15:54 EDT
References: <426@oakhill.UUCP> <8745@microsoft.UUCP>
Organization: Charles River Data Systems, Framingham MA
Lines: 31
Xref: watmath net.arch:1332 net.lang.c:5355 net.micro:10693 net.micro.pc:4168 net.micro.68k:880

I ran the Knight's Tour program on CRDS' current 68000 computer, and on
our prototype 68020.  I also compared our old C compiler against our new
C compiler from Green Hills.  For comparison, I also ran it on a VAX 750
and a PDP 11/70 (my pet system at MIT).  Also, for fun, I ran both 5x5 and
6x6 boards.  The results:

	12.5Mc	[1]	12.5Mc	[2]	12.5Mc		Cast of Compilers
Board	68000	VAX750	68000	11/70	68020		WS C	Whitesmiths
Size	WS C	 pcc	GH C	V7 C	GH C		GH C	GreenHills
	-----	-----	-----	-----	-----		pcc	pcc VAX
5 x 5	  5.1	  3.5	  3.5	  2.3	  1.7		V7 C	God's own C
6 x 6	147	109	 61	 67 	 29.5
7 x 7		       2859*				* Ran on "frog", our
							uucp machine, in
[1] mit-eddie						background, just for
[2] mitccc						amusement.

All these machines were lightly loaded during the tests, the times reported
are user-times (in seconds).

The 680?0 C compilers both use 32-bit "int"s.

One advantage the Greenhills compiler has over the old one is that it tries
to make clever use of spare registers lying around.  I found the difference
quite instructive.

-- 
John Woods, Charles River Data Systems, Framingham MA, (617) 626-1101
...!decvax!frog!john, ...!mit-eddie!jfw, jfw%mit-ccc@MIT-XX.ARPA

"MU" said the Sacred Chao...