Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!unix.cis.pitt.edu!pitt!vax.cs.pitt.edu!jonathan
From: jonathan@cs.pitt.edu (Jonathan Eunice)
Newsgroups: comp.sys.hp
Subject: Re: HP 9000 s 700 questions ????
Message-ID: <JONATHAN.91Mar30172045@speedy.cs.pitt.edu>
Date: 30 Mar 91 22:20:45 GMT
References: <1991Mar13.121715.25269@qut.edu.au> <31480011@hpcuhe.cup.hp.com>
	<1991Mar16.014605.11367@mcs.kent.edu> <7401@titcce.cc.titech.ac.jp>
	<1991Mar19.221804.22673@riacs.edu> <1995@kuling.UUCP>
Sender: news@pitt.UUCP
Organization: University of Pittsburgh Computer Science
Lines: 22
In-reply-to: irf@kuling.UUCP's message of 26 Mar 91 08:12:50 GMT

Bo Thide' writes:


   >Now that the 700 series has been announced, can someone post or
   >mail me the Linpack MFLOP ratings for each model.

   HP9000/720 (Cobra):  Linpack Double Precision: 17 MFLOPS
   HP9000/730 (King Cobra):  Linpack Double Precision: 22 MFLOPS
   HP9000/750 (Coral):  Linpack Double Precision: 22 MFLOPS

   This is for HP-UX 8.05.  For HP-UX 8.01 these figures may be a
   bit lower (15 MFLOPS for the 720, I think).  Still, the fastest
   FP workstation around.

Hardly.  The IBM RS/6000 Models 550 and 950 give 25.2 MFLOPS, or
79.0 SPECfp's.  The Stardent 750 gives 25 MFLOPS on vectorizable
code, such as Linpack (SPECfp unknown to me).

HP *has* done exceptionally well, but hasn't trounced those really 
concentrating on FP performance.  On a price/performance basis, on
the other hand, it is banging hard on IBM, Stardent, etc., especially
on the low end.