I've done similar tests and found similar results. But more high
performance workstations such as the Sun Ultra 60 with 1 gig of ram, blow
away the Intel Linux on my 400Mhz PC and my Ultra 10. I've found it to
be faster by a factor of 3 than either machine, which in my tests
performed quite similarly. Of course, it's also twice as expensive as
the Ultra 10.
Edward Kovalik
[log in to unmask]
UCLA Division of Brain Mapping
http://www.brainmapping.org
(310) 206-4323
WLA-VA PET Research Facility
(310) 478-3711, x. 49018
On Tue, 2 Feb 1999, Gregory S. Berns wrote:
> Giuseppe Pagnoni wrote:
> >
> > Dear spm list
> >
> > we are thinking of buying a high performance PC (450 MHz, 512 Mb of RAM)
> > implementing linux to run spm. Does anybody know how such a choice would
> > compare in terms of efficiency and speed (in terms of costs it's quite
> > clear...) with respect to a commonly used workstation (SUN, SGI) for SPM
> > processing?
> >
> > thanks!
> >
> > bests
> > Giuseppe
>
> We did a head-to-head comparison of:
> PC (PII 350 MHz, 256 MB RAM, running FreeBSD)
> vs.
> Sun Ultra 10 (320 MB RAM)
>
> doing a typical SPM of a PET dataset.
>
> Guess what? The PC was faster by a factor of 2.
> Note that you will have to do a bit of hacking to get matlab and spm
> working properly under Linux, though several listmembers have done so.
> We went with FreeBSD because it's apparently a bit faster for
> networking. FreeBSD has a perfectly good Linux emulator, and I
> run the Linux version of Matlab with all the mexlx files for spm.
>
> Greg Berns
> Emory University School of Medicine
>
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|