🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: comp.sys.dec
Article #351

Re: tps benchmarks for VAX 3900

#351
From: Amitabh Shah
Date: Thu, 09 Apr 1992 14:41
51 lines
3285 bytes
On 8 Apr 92 16:36:49 GMT, amf@amfent.Gwinnett.COM (Andy Feibus) said:
> Xref: nntpd.lkg.dec.com comp.sys.dec:9703 comp.os.vms:51727

>> On 6 Apr 92 16:56:42 GMT, mdchaney@bronze.ucs.indiana.edu (M Darrin Chaney) s
>>> Does anyone have the TPPC results for a VAX 3900.  I'm particularly
>>> interested in the tpsA-Local.

>> No, there are no TPC-A results announced on the VAX 3900. Nor are they likely
>> to be announced. The closest machine that DEC has annouced TPC-A results for 
>> the MV 3100-30, which is 21.6 tpsA-local (@$8.6K/tps). 3900 is rated at aroun
>> 4 vups, while the 3100-30 is rated at around 5 vups. You make the final
>> extrapolation.
 
> Arrgg!  Bad advice, Amitabh.

> You cannot extrapolate TPC-A numbers for one system by using another
> system.  He'll have just as much luck trying to extrapolate TPC-A numbers
> for the 3100-30 by using a different DBMS.  Or an RS-6000.  It just doesn't 
> work that way.

I disagree with you. People extrapolate results all the time. Just peek over
into comp.benchmarks. As a Digital representative to the TPC, I know enough
that I can not publicly extrapolate any TPC results. That is why I left the
final step to Darrin. I do not even suggest whether the extrapolation is
linear or non-linear, etc. 

OTOH, I actually have an estimated tpcA-local number for the 3900. It is
extrapolated from various pieces of data, but it is outdated (we have had
newer software releases, better understanding of TPC-A benchmark, etc). But I
would never disclose it publicly, since it is disallowed by the TPC. But this
restriction is only on the TPC members. The customers of these vendors are
free to estimate and extrapolate as much as they want. Indeed, this will be
the case for systems such as the 3900 which is no longer made by DEC, and
hence will never get an official tpc-a number. 

One final note. While TPC-A is a good system level benchmark, it is not
perfect. TPC-C, the benchmark that the TPC is developing, and is expected to
be released soon, will be a better system level benchmark than TPC-A. But,
even TPC-C will not be perfect. Finally, the customer will have to look at all
the available data, use their own benchmarks, and use judicious extrapolation
to consider their final needs.

-amitabh.

--
______________________________________________________________________________
Amitabh Shah                                   shah@tay1.dec.com----(INTERNET)
Digital Equipment Corporation                  santur::shah----------(EASYNET)
Software Performance Group                     +1 508 952 4430---------(PHONE)
151 Taylor St. Littleton MA 01460              +1 508 952 4197-----------(FAX)
Technical Details
Message-ID: <SHAH.92Apr9104111@santur.tay1.dec.com>
Path: arc1-ca.pugleaf.net!sparky!uunet!decwrl!pa.dec.com!nntpd.lkg.dec.com!shah
References: <SHAH.92Apr6164054@santur.tay1.dec.com> <3Z8wiB2w164w@amfent.Gwinnett.COM>

Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!decwrl!pa.dec.com!nntpd.lkg.dec.com!shah
From: shah@santur.tay1.dec.com (Amitabh Shah)
Newsgroups: comp.sys.dec,comp.os.vms
Subject: Re: tps benchmarks for VAX 3900
Message-ID: <SHAH.92Apr9104111@santur.tay1.dec.com>
Date: 9 Apr 92 14:41:11 GMT
References: <SHAH.92Apr6164054@santur.tay1.dec.com> <3Z8wiB2w164w@amfent.Gwinnett.COM>
Sender: news@nntpd.lkg.dec.com
Followup-To: comp.sys.dec
Organization: Digital Equipment Corporation, TAY1, Littleton MA
Lines: 52
In-reply-to: amf@amfent.Gwinnett.COM's message of 8 Apr 92 16:36:49 GMT