home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.arch:10670 comp.benchmarks:1675
- Path: sparky!uunet!ogicse!das-news.harvard.edu!cantaloupe.srv.cs.cmu.edu!lindsay
- From: lindsay+@cs.cmu.edu (Donald Lindsay)
- Newsgroups: comp.arch,comp.benchmarks
- Subject: Re: DEC ALPHA Performance Claims
- Message-ID: <BxME1C.6Gt.2@cs.cmu.edu>
- Date: 12 Nov 92 20:24:44 GMT
- Article-I.D.: cs.BxME1C.6Gt.2
- References: <1992Nov11.043149.17257@engage.pko.dec.com> <BxKI38.DM7.2@cs.cmu.edu> <1992Nov12.091854.22914@walter.cray.com>
- Sender: news@cs.cmu.edu (Usenet News System)
- Organization: School of Computer Science, Carnegie Mellon
- Lines: 25
- Nntp-Posting-Host: gandalf.cs.cmu.edu
-
-
- cmg@magnet.cray.com writes:
- > (NOTE: The performance above which is attributed to a CRAY-1 is
- > actually that of a CRAY S-MP, NOT a CRAY-1.
-
- Yes, I screwed up. Sorry it bothered you: I hadn't posted a
- correction, because that number wasn't important to my point.
-
- >On this particular benchmark,
- >we would see a 3x speed improvement attributable to software.
- >The improvement of microprocessor speed of the past decade has been
- >tremendous. How much of this improvement is attributable to software?
-
- We should get John Mashey to give us a graph of SPECmarks versus
- time, chip held constant. Several chips (the 88100, for instance)
- noticeably picked up speed this way.
-
- In general, optimization has the most leverage on numerical
- benchmarks, since their execution times are most likely dominated by
- some smallish stretch of code. We would expect SPECint to have been
- bent less than SPECfp. Of course, in real life, a lot of performance
- is dominated by disk speeds and the like, so improved compiler
- technology doesn't always play a role.
- --
- Don D.C.Lindsay Carnegie Mellon Computer Science
-