Newsgroups: comp.sys.apollo
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!caen!news
From: ericb@srvr1 (Eric Bratton,235 Chry Ctr,9363561,)
Subject: Re:  Where is the 68040
Message-ID: <1991Mar18.185632.2080@engin.umich.edu>
Sender: news@engin.umich.edu (CAEN Netnews)
Organization: The University of Michigan, Ann Arbor
References: <9103181736.AA16392@richter.mit.edu>
Date: Mon, 18 Mar 1991 18:56:32 GMT

From article <9103181736.AA16392@richter.mit.edu>, by krowitz@RICHTER.MIT.EDU (David Krowitz):
> If you haven't already noticed in the latest issue of HP's FYI
> mailing, they've reduced the Mflop rating of the 68040. When
> they first announced the 425t last June, HP quoted 20 MIPS and
> 3.5 Mflops. Now they are saying 20 MIPS and 2.6 Mflops, or
> 2.9 Mflops if you have an optional external cache installed
> to supplement the on-chip data cache. I don't remember if the cache
> can be added to the desktop boxes (ie. 400t/425t), it may only
> be available on the deskside boxes (ie. 4002/425s/433s). 

We have a 425T now, and according to the ADUS linpackd.ftn, they
run at 2.2 MFLOPS.  This is with the Rev 10.8 fortran compiler with
the "-opt 4 -cpu mathlib" flags.  A DN3500 returns with about .25 MFLOPS,
for comparison.  Your milage may vary, of course.

--
*-----------------------------------------------------------------*
| Eric Bratton                | The University of Michigan
| ericb@caen.engin.umich.edu  | #include <std.disclaimers>
*
