Path: utzoo!utgpu!watserv1!watmath!att!rutgers!cs.utexas.edu!uunet!mcsun!unido!ira.uka.de!fauern!tumuc!lan!charly.bl.physik.tu-muenchen.de!k2
From: k2@charly.bl.physik.tu-muenchen.de (Klaus Steinberger)
Newsgroups: comp.sys.m68k
Subject: Re: Moto's data predicts 68040 performance well below 20 MIPS
Message-ID: <3445@tuminfo1.lan.informatik.tu-muenchen.dbp.de>
Date: 16 Jul 90 07:06:25 GMT
References: <40088@mips.mips.COM>
Sender: news@lan.informatik.tu-muenchen.dbp.de
Lines: 41

mark@mips.COM (Mark G. Johnson) writes:

>So, the data and the claim that 68040==20VAXmips  implies that the earlier
>68020 has a "sustained performance level of 4.9 VAX-equivalent MIPS"
>(4.9 = 20/4.1).  Does anybody seriously believe this?

I don't know which clock-rates are used in this test, but
on 25 Mhz or 33 Mhz, I think it's possible, but only for Integer!!!!!


>About the most impartial data I could find was for the Hewlett Packard
>HP9000 model 370 machine.  This uses a 68030 (not 68020) at 33 MHz (not
>25 MHz) and achieves a geometric mean of 3.9 SPECmarks [ref. SPEC
>newsletter v1#1].  It seems reasonable to suspect the 68020 is no
>better than the 030 in performance {else who'd want the 030?}, so we
>conclude that the 020's performance is, at most, 3.9 VAX-equivalent

Don't mix up SPECmarks and MIPS, SPECmarks measures also the 
Floating point power. To our experience, the VAX 780 has nearly the same 
floating point speed than a 68881. (Most times the FP is even faster,
due to better compilers)

If the data from Motorola is correct, the '040 Floating Point will
be more than a magnitude faster than the 68881. (3.5 Mflops Linpack
compared to 0.1 Mflops). I don't have Linpack data for the HP9000/370,
but a Tektronix4315 (w. 68882) is claimed to 0.17 Mflops.
Linpack data for a Vax11/780 is .14 Mflops.

I don't know if Motorola has used all Fortran, or coded BLAS for the
Linpack, but this should make a difference of at most 30 %.

>Of course the best method would be to lay hands on an actual computer
>system that uses the 68040 and benchmark it; presumably Motorola
That's really the best method.

Sincerely,
Klaus Steinberger

Klaus Steinberger               Beschleunigerlabor der TU und LMU Muenchen
Phone: (+49 89)3209 4287        Hochschulgelaende, D-8046 Garching, West Germany
BITNET:  K2@DGABLG5P            Internet: k2@charly.bl.physik.tu-muenchen.de