Are SPECfp double precision?

Thread View: aus.sun-works

2 messages

[T] Back to Threads [V] Tree View [G] Group Articles

2 total messages Started by Mathew BM LIM Wed, 25 Nov 1992 23:08

#49

Author: Mathew BM LIM
Date: Wed, 25 Nov 1992 23:08

9 lines
876 bytes

The subject says it all, I want to know if SPECfp performance figures are 
done using double or single precision numbers.


-- 
Mathew Lim, Unix Systems Programmer, ANU Supercomputer Facility,	
Australian National University,	GPO Box 4, Canberra City, ACT, Australia 2601.
Telephone : +61 6 249 2750 	| 
Fax       : +61 6 247 3425	| E-Mail  : M.Lim@anu.edu.au

Re: Are SPECfp double precision?

#50

Author: Gregory N. Bond
Date: Thu, 26 Nov 1992 22:23

24 lines
1775 bytes

>>>>> On 25 Nov 1992 23:08:34 GMT, mbl900@anusf (Mathew BM LIM) said:

>> The subject says it all, I want to know if SPECfp performance figures are 
>> done using double or single precision numbers.

I believe the SPEC programs are all written in C, which would imply
double precision.

The difficulty with the SPECfp, (and the reason SPEC was split into
the int and fp sections), is that fp is a pretty diverse area, parts
of which are amenable to tweaks.  The classic example is the RS6000
architecture, that runs some of the SPECfp stuff blindingly fast
(because the problems match the architecture well), but some only
so-so.  Just as important as the "raw" SPECfp figure is the variation
on each test, all of which must be made available (as part of the
licence for using the SPEC name); interpreting this in the light of
your particular problem set is at best obscure.

Greg.


--
Gregory Bond <gnb@bby.com.au> Burdett Buckeridge & Young Ltd Melbourne Australia
``There is Faith, Hope and Charity.  But greater than these is Banking.'' - 1492

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads