home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.benchmarks:1680 comp.arch:10697
- Newsgroups: comp.benchmarks,comp.arch
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!decwrl!pa.dec.com!nntpd2.cxo.dec.com!nntpd.lkg.dec.com!ryn.mro4.dec.com!wrksys.enet.dec.com!bhandarkar
- From: bhandarkar@wrksys.enet.dec.com (Dileep Bhandarkar)
- Subject: Re: DEC ALPHA Performance Claims
- Message-ID: <1992Nov13.153624.28222@ryn.mro4.dec.com>
- Sender: news@ryn.mro4.dec.com (USENET News System)
- Organization: Digital Equipment Corporation
- References: <BxH7s7.5Cv@inews.Intel.COM> <1992Nov10.150330.26923@ryn.mro4.dec.com> <1992Nov10.181828.24622@sol.UVic.CA> <BxJ47t.K65.2@cs.cmu.edu> <1992Nov12.193431.13848@sol.UVic.CA>
- Date: 13 NOV 92 10:35:44
- Lines: 17
-
-
- In article <1992Nov12.193431.13848@sol.UVic.CA>, zapantis@uvphys.phys.uvic.ca (Nik Zapantis) writes...
- >
- >DEC has been quoting Mflops for a number of their systems, such as the
- >DECstation 5000 family...etc. I am asking if they can supply a Mflop rating
- >for the Alpha, that is obtained by using the same Mflop benchmarks that they
- >have used to rate some of their other systems. For example, the VAX9000VP
- >is rated at 125Mflops. How does the Alpha compare to the VAX9000VP?
- >
- The 125 MFlops is a peak rate assuming a vector multiply and a vector add
- simultanously at a 16 ns (62.5 MHz) cycle time. For the current Alpha AXP
- family, the peak megaflops are same as clock rate i.e. 133 to 200 Mflops.
-
- Peak megaflops are "guaranteed to never exceed" numbers. Megaflops for various
- sizes of Linpack, Livermore Loops, or Perfect Club reflect actual delivered
- performance in some benchmark environment. I did post these a couple of days
- ago.
-