HP3000-L Archives

February 1995, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Eero Laurila <[log in to unmask]>
Reply To:
Eero Laurila <[log in to unmask]>
Date:
Tue, 14 Feb 1995 00:04:24 GMT
Content-Type:
text/plain
Parts/Attachments:
text/plain (42 lines)
Chris Bartram ([log in to unmask]) wrote:
:  In <[log in to unmask]> Eero Laurila <[log in to unmask]> writes:
: > Performance figures compared to a series 42:
: >   Micro3000  42(=48)    42XP/52     s70          s947
: >   1.0        1.0        1.4         3.7          5.2
 
: Eero,
:   Do you have a continuing chart of relative performance - say from the Micro
: up to the new 99x boxes? I've never seen a *complete* performance chart (even
: rough estimates would be useful) and would be nice to add to the FAQ...
 
Sorry to say but no, I don't.  Those figures came out from about 3..4 years
old information, which at that time was considered confidential.  I figured
that those models were old enough to pass out above numbers as 4 out of five
were close to 10 years old.
 
Another thing is that the numbers above were related to plain CPU power only,
and were not adjusted to take into account different i/o buses/structures/
throughputs.  I.e. they cannot be used for direct system performance
comparison - a system with faster central buses/io-buses/more channels/better
design gives better system performance even if CPU torque number was lower
or same...
 
For current CPU's performance figures, I'm afraid it's confidential and
as such even if I knew I would not be able to release them without HP's
blessing.  All I have for current CPU's is what's published in HP's
marketing documents.  Those typically use a s950 as a reference machine
and it's relative performance is considered to be 1.0.
 
From HP's brochyres I can roughly figure out what the ratios are but that's
probably not accurate and might reflect the systems relative performance
including it's i/o architecture (or not(?)).  Anyway, using those brochyres,
I could roughly say that series 37 is the slowest (I don't have series II, III,
series 30/33 info anywhere - wonder if anyone else has - compared to something
later than 37 (from 1984))...  and 995-800 fastest.  Range being about 150
times the performance of a 3000/37.  I'd be tempted to say it's even more...
four users on a 37 would probably not get better response time than 2000
on 8-way Emerald!!  But as said.. these are only estimates based on
marketing docs...
 
Eero - HP CSY lab.

ATOM RSS1 RSS2