From owner-freebsd-alpha  Sun Dec 13 03:58:34 1998
Return-Path: <owner-freebsd-alpha@FreeBSD.ORG>
Received: (from majordom@localhost)
          by hub.freebsd.org (8.8.8/8.8.8) id DAA20524
          for freebsd-alpha-outgoing; Sun, 13 Dec 1998 03:58:34 -0800 (PST)
          (envelope-from owner-freebsd-alpha@FreeBSD.ORG)
Received: (from jmb@localhost)
          by hub.freebsd.org (8.8.8/8.8.8) id DAA20421;
          Sun, 13 Dec 1998 03:58:00 -0800 (PST)
          (envelope-from jmb)
Date: Sun, 13 Dec 1998 03:58:00 -0800 (PST)
Message-Id: <199812131158.DAA20421@hub.freebsd.org>
From: "Jonathan M. Bresler" <jmb@FreeBSD.ORG>
To: jim@cgi.sstar.com
Cc: alpha@FreeBSD.ORG, hardware@FreeBSD.ORG
In-reply-to: <199812112343.RAA15525@cgi.sstar.com> (message from Jim King on
	Fri, 11 Dec 1998 17:43:59 -0600 (CST))
Subject: Re: Hint benchmark on Multia VX40A
References:  <199812112343.RAA15525@cgi.sstar.com>
Sender: owner-freebsd-alpha@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org


Thanks to Andrew Gallatin for running the Hint benchmarks on a Digital
(DEC now Compaq) Alpha Multia VX40A.

integer performance is nearly the same as an AMD Am5x86-133.  that
places it just a tad slower than a Pentium 90 until the dataset size
reaches about 12kB (then the p90 falls out of cache).  at that point
the p90 the am5x86-133 and hte multia are identical till 250kB when
the am5x86-133 starts lagging behind the other two.

double performance is half way between a Pentium 90 and a PentiumPro
150.

conclusions:  if you have TrueParity (9bit) memory to spare or dont
mind playing three times the price of common memory AND you have a lot
of floating point work to do, the multias may be a good fit.  as a
general purpose computer, they are woefully underpowered.

jmb

ps.	hopefully that creates a record in the mailing lists that will
        help anyone using the search engine to find out about
        multias. ;)

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-alpha" in the body of the message