home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!usc!rpi!uwm.edu!ogicse!das-news.harvard.edu!cantaloupe.srv.cs.cmu.edu!lindsay
- From: lindsay+@cs.cmu.edu (Donald Lindsay)
- Newsgroups: comp.arch
- Subject: Re: DEC Alpha AXP System INTEGER Performance
- Keywords: Alpha, AXP, SPEC, DEC, INTEGER
- Message-ID: <By4w36.EMo.2@cs.cmu.edu>
- Date: 22 Nov 92 20:11:30 GMT
- Article-I.D.: cs.By4w36.EMo.2
- References: <1698@niktow.canisius.edu> <jdd.721687838@cdf.toronto.edu> <1992Nov20.220615.7494@raid.dell.com>
- Sender: news@cs.cmu.edu (Usenet News System)
- Organization: School of Computer Science, Carnegie Mellon
- Lines: 21
- Nntp-Posting-Host: gandalf.cs.cmu.edu
-
- samf@yosemite.dell.com (Sam Fuller) writes:
- >After reviewing the Alpha performance claims, I had the feeling that
- >the integer unit did not really run at 133Mhz.
-
- No, it runs full speed.
-
- I tried a worst-case benchmark: a program which
- - uses no FP, hence, the superscalar doesn't apply much
- - manipulates a byte array (an image)
- - makes multiple passes over an array bigger than the secondary cache
-
- A 150 MHz OSF/1 Alpha was ~4.4 times faster than a 25 MHz R3000
- (DECstation 5000/200). By comparision, the clocks are 6:1, the DRAM
- speed is probably about 1:1, and your guess would have predicted 3:1.
-
- A quick look at the generated code shows room for improvement, and I
- expect that future compiler releases will do better. For one thing,
- the compiler's profiling and feedback features aren't available to me
- yet.
- --
- Don D.C.Lindsay Carnegie Mellon Computer Science
-