[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: MCL and PowerBooks
- To: straz (Steve Strassmann), Narinder Singh <singh@Hudson.Stanford.EDU>, email@example.com
- Subject: Re: MCL and PowerBooks
- From: firstname.lastname@example.org (Kalman Reti)
- Date: Wed, 7 Sep 1994 17:53:03 -0400
At 2:10 PM 9/7/94 -0400, Steve Strassmann wrote:
>At 11:29 AM 9/6/94 -0700, Narinder Singh wrote:
> >I am looking for some advice/comments on PowerBooks for MCL. The only
> >reason we have bought PowerBooks in our group is to run MCL.
> >We currently have some PowerBook 180cs and we are considering the
> >PowerBook 540c and the PowerBook Duo 280c. I have read the official
> >specs to find out the obvious hardware differences. I am interested in
> >finding out:
> > 1) Does MCL run on the 540c and 280c?
> > 2) How do the 540c and 280c compare with the 180c in speed (especially for
> >floating point operations)?
> > 3) any other comments/experiences that are relevant to MCL
>MCL should run on the 540c ok if you add the missing-FPU patch.
>Get it from cambridge.apple.com:/pub/mcl/patches/missing-FPU/
>In any case, you should upgrade to MCL 2.0.1. The upgrade is in
>this same directory, or you can order it from APDA for $25
>(includes a cdrom and three diskettes).
>I'm afraid I have no benchmark info.
At the request of another customer, I tried an exceedingly simple
benchmark on a 520c (which has the same speed processor as the
540c). The times went from 3.1-3.2 seconds on a Quadra 950 to
23.6-23.8 seconds on the 280c, or more than 7.6 times slower. (Oddly
enough, the same thing on a IIfx takes only 3.2-3.5 seconds, suggesting
that this might not be such a good benchmark.)
The mail from the other customer suggesting the "benchmark" is below.
I'd be very happy to get more representative benchmarks and publish
>I don't have a "standard" benchmark for floating-point speed,
>but I have just dreamt up the following simple test:
>(defun foo ()
> (dotimes (i 10000)
> (* (expt pi (- 1.23456789 (/ (+ i 1.23456789))))
> (atan (sin (cos i))))))
> (dotimes (n 10)
> (time (foo))))
>I put the above expressions in a small file, and then started up a
>brand new copy of MCL2.0.1, ... EXCEPT THAT I FIRST INCREASED ITS MEMORY
>ALLOCATION IN THE FINDER to 6144 K.
>Loading the source file into MCL gives the following results.
>(I tried it on both my PB180C and my old Mac IIfx ...quite a difference
>in speed... good old IIfx. The Powerbook is hovering around 6.4 seconds,
>and the IIfx around 3.5 seconds, total times).