Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uunet!wuarchive!cs.utexas.edu!oakhill!davet
From: davet@oakhill.UUCP (David Trissel)
Newsgroups: comp.arch
Subject: Re: MC68040 benchmark claims
Message-ID: <2942@oakhill.UUCP>
Date: 12 Feb 90 08:20:48 GMT
References: <8859@portia.Stanford.EDU> <5190@convex.convex.com> <1850@cbnewsi.ATT.COM> <2938@oakhill.UUCP> <35816@mips.mips.COM>
Reply-To: davet@oakhill.UUCP (David Trissel)
Organization: Motorola Inc., Austin, Texas
Lines: 15

In article <35816@mips.mips.COM> crisp@mips.COM (Richard Crisp) writes:

>> is 23,148 KDhrys. The MC68040 runs that benchmark at twice the speed.
>
>Dave, would you describe the environment (hardware, software, clock rate,
>memory access time (in clock cycles)) for the 68040 system whose performance
>you quote? As has been stated time and again in this forum, without such
>data benchmark results are meaningless.

The benchmark time is for a 25 Mhz MC68040. The software is the MC68UCC4.2 C
compiler with -O and #include string.h (inline strings.) External memory 
times have no effect on the execution time of Dhrystone on any MC68040 system. 
This is because the benchmark runs entirely out of the MC68040 on-chip cache.

 -- Dave Trissel - Motorola Semiconductor, Austin