home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.mac.misc
- Path: sparky!uunet!stanford.edu!CSD-NewsHost.Stanford.EDU!Xenon.Stanford.EDU!torrie
- From: torrie@cs.stanford.edu (Evan Torrie)
- Subject: Re: Macs cost too much (NOT!)
- Message-ID: <torrie.712476567@Xenon.Stanford.EDU>
- Originator: torrie@Xenon.Stanford.EDU
- Sender: news@CSD-NewsHost.Stanford.EDU
- Organization: CS Department, Stanford University, California, USA
- References: <D2150035.7sh4j4@outpost.SF-Bay.org> <ajross.711133468@husc9> <1992Jul15.115921.14033@msc.cornell.edu> <ajross.711328520@husc9> <5794.2a6ac834@hayes.com> <92203.173612ASI509@DJUKFA11.BITNET> <5805.2a6d64c8@hayes.com> <92205.170150ASI509@DJUKFA11.BIT <92211.150235ASI509@DJUKFA11.BITNET>
- Date: 30 Jul 92 06:09:27 GMT
- Lines: 44
-
- ASI509@DJUKFA11.BITNET writes:
-
- >The FIRTS benchmark I referred to in my post was a Dhrystone benchmark
- >compiled with THINK C 4.05 on the Mac IIsi and GCC 2.1 running under OS/2 2.0
- >on a 33MHz 386 (NOT 486 that will be another factor 2 to 2.5). The results:
-
- Let's just say that the difference in code quality between Think C 4.0.5
- and GCC 2.1 is LARGE (and it's in GCC 2.1's favour). Maybe if you had
- done a test using GCC on the Mac, you would have a valid comparison.
- However, see below.
-
- >Now you can discuss the meaningfullness or lack thereof of this kind of bench-
- >mark, but since this test comes from the UNIX (or mainframe ?) world and most
-
- If you knew a bit more about the benchmark you're running, you'd know
- that Dhrystone is essentially DEAD as a benchmark in the Unix world.
- It has been thoroughly discredited as prone to weird optimisations
- which make little sense in real code. See Hennessy and Patterson
- "Computer Architecture - A Quantitative Approach" for details.
-
- >of the machines mentioned in the source code were 680x0 based workstations I
- >seriously doubt it is highly optimized for the PC.
-
- Actually, Dhrystone is one benchmark which happens to optimise particularly
- well on 80x86 architectures, because it has a large percentage of
- string code, which maps well to the 80x86 instruction set.
- I can find you synthetic benchmarks which show the Mac is much
- faster than an 80x86. Try Savage for example.
-
- >The SECOND "benchmark" I ran was using Word 4.0 on the Mac and WinWord 1.1 on
- >the same PC with Win3.0 (not the fastest GUI of all times). The PC was equipped
- >with a ET3000 VGA board which is definitely not the fastest under the SUN :-)
- >and in fact already out of production. Both Word versions are functionally
- >equivalent and --- you might guess it --- the PC won this subjective
- >"benchmark" hands down. It is _my_ Mac and _my_ PC so why should I be that much
- >biased ?.
-
- So you're basing your views on Microsoft code? Somehow, that doesn't
- sound like something I'd do.
-
- --
- ------------------------------------------------------------------------------
- Evan Torrie. Stanford University, Class of 199? torrie@cs.stanford.edu
- Embrace rationalism, reject superstition. Break away from the past.
-