SUMMARY: What is a SPECrate, really?

From: SJ Higa (shiga@nexus.aer.loral.com)
Date: Fri Feb 17 1995 - 14:23:52 CST


ORIGINAL POST:

> I've been asked by my manager to explain what a SPECrate is. All
> I know is that it's a widely used benchmark that applies to multi-
> processing systems and that as usual with benchmarks "Bigger is
> Better". But what in fact is a SPECrate? I'm hoping someone out
> there can enlighten me as to what this thing actually measures
> and what factors (CPU speed, disk access, etc.) are used to produce
> this supposedly meaningful single number.

SUMMARY:

Thanks to all respondents, especially to Victor Fieldhouse for sending
me a copy of the SPEC FAQ. This can be retrieved by anonymous ftp
(file://ftp.nosc.mil/pub/aburto/spec.faq). Anyone wanting information
on SPEC would do well to begin with the FAQ. Basically, the SPECrate
int and fp are produced from a suite of benchmarks which are
themselves measures of CPU, memory, and compiler code generation
performance. Disk and other I/O effects are negligible factors.

THANKS TO:

victor@estwns.wm.estec.esa.nl (Victor Fieldhouse)
asargusi@sbrsim.ed.dreo.dnd.ca (Ayrton Sargusingh)
Kevin.Sheehan@uniq.com.au (Kevin Sheehan {Consulting Poster Child})
stern@sunrise.East.Sun.COM (Hal Stern - NE Area Systems Engineer)
misawa@physics.Berkeley.EDU (Shigeki Misawa)

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Stan Higa
Loral Aeronutronic
29947 Avenida De Las Banderas Email: shiga@nexus.aer.loral.com
P.O. Box 7004
Rancho Santa Margarita, CA 92688
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:16 CDT