Xref: utzoo comp.unix.i386:3576 comp.unix.xenix:10613
Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!milano!uudell!loft386!dpi
From: dpi@loft386.UUCP (Doug Ingraham)
Newsgroups: comp.unix.i386,comp.unix.xenix
Subject: Re: BENCHMARKS (Byte's) 368 vs 486 comp.
Summary: Old benchmark, results not the same.
Message-ID: <618@loft386.UUCP>
Date: 15 Mar 90 18:01:26 GMT
References: <1990Mar8.175245.2527@pcrat.uucp> <1025@fiver.UUCP> <18136@rpp386.cactus.org>
Followup-To: comp.unix.i386
Organization: Lofty Pursuits, Rapid City, SD USA
Lines: 25

In article <18136@rpp386.cactus.org>, jfh@rpp386.cactus.org (John F. Haugh II) writes:
> I agree.  At the 1988 USENIX Winter conference in Dallas [ I think that
> is the right year - it's the one I went to in Dallas ;-) ] there was a
> 20 or 25 MHz Compaq using the Greenhill's C compiler producing about 9KDhry.
> 
> The Greenhills compilers have always impressed me and I do plainly recall
> that this particular Compaq was running about 2KDhry faster than what
> everyone else was showing.

This must have been with the old dhrystone benchmark.  I remember getting
amazing numbers with that compiler on that test.  The old dhry benchmark
had a couple of problems that allowed portions of the test to be optimized
away if the compiler was clever enough.  The Greenhills compiler had the
ability to do that particular optimization.  As it turns out most programs
don't have enough useless code that it makes a difference.  The Greenhills
compiler just wouldn't compile some programs without lots of diddling with
source.  I decided not to use it because of this.

I will see if I can find a copy and run the new dhry2 with the compilers
I have at hand if there is any interest.

-- 
Doug Ingraham (SysAdmin)
Lofty Pursuits (Public Access for Rapid City SD USA)
uunet!loft386!dpi