I have an old HP3000 Processor Selection Guide (HP Part #5953-3738)
dated 4/89 proudly hanging up in my office. The highest processor HP
then had was a 955, rated at approx 4.5. The closest IBM 3090 offering
you refer to then was a 3090-500S rated at approximately 75. These were
OLTP ratings. I don't know if HP still maintains these charts - I doubt
it with the rapid rate of change, however I am sure someone in HP would
have this competitive info.

At worst case you may be able to extrapolate from these figures eg.
when we upgraded from a 955 to 987 we used a figure of 3.2 times performance.

All of these numbers are 'rubbery' and depend entirely upon your
environment and the software you are using eg. we were using a 3000/987
to run MRP (Manman). We are now using SAP R/3 on 'big iron' HP9000's
(many). Our Backflush process has blown out from 1 - 2 secs. to (in
some cases) 80 - 90 secs, MRP from 15 minutes to 6 hours!. The
performance of the combined new system we are running on is probably an
order of magnitude more than our trusty 987. There are obiously a lot
more issues with our blowouts than the simplistic picture painted here,
nonetheless one should not get too carried away with relative
performances of machines without taking into account all of the other issues.

Wayne Laughton
Southcorp Water Heaters Australia
[log in to unmask]
Phone   + 61 2 9684 9140
Fax     + 61 2  684 4270
Alan St. Rydalmere  NSW  2116
Australia
----------
From: HP3000-L; john.schick
To: HP3000-L
Subject: HP3000 - IBM System Comparisons
Date: Friday, 10 January 1997 4:06

Our shop runs an HP3000 Series 995/400 with MPE/iX 5.0, 704 MB memory
and 42 GB disk storage.  We're supporting an average of 350
interactive sessions running mostly COBOL/TurboImage/VPlus
applications.  We've been acquired by a company that runs "big iron"
from "big blue" (IBM 3090-500J systems running MVS to be precise).
The new management is asking for performance ratings on the HP3000
995/400 that relate to systems with which they are familiar.

Can anyone provide me with a rough estimate comparing the system
performance of an HP3000 995/400 with an IBM 3090-500J?

What IBM system would most closely compare in overall performance to
the HP3000 995/400?

I realize it's difficult (if not impossible) to be very precise
without details of the entire application environments, but any
general thoughts/opinions/experiences on this subject would be
appreciated.

John Schick  ([log in to unmask])
Manager/Technical Support
McGraw-Hill Higher Education