Linux and bioinformatics questions....

[Chris Botka]

Francois Jeanmougin wrote on 2 Aug 1999:
> In article <37A57241.BAF5EE2B at micromet.de>,
> Thomas Busl <thomas.busl at micromet.de> writes:
> > Dear Tim,
> > I work with solaris on sparc and I think you are right about SMP
> > workstations. You can see that sun has alot of expiriences in
> > that field. I think you get solaris2.6 free for intel plattforms.
> > so long
>
> Actually, with big databases, the problem is not Linux vs Solaris
> (Because, Linux runs well on a Sparc SMP), but 32 vs 64 bits
> processors. And in this field, it could be good to test a Alpha.
> The "old" 21164PC runs at 533MHz and can handle 128bits bus at
> 100MHz. For a low price 64bits calculator, you could consider
> making a beowulf linux clusteer of such machines.
>
> Francois.

I know this thread is a bit stale, but I must second the suggestion  
to test an Alpha.  We recently upgraded to the 21264 alphaserver  
ES40.  What was a very fast machine (alphaserver 4100/21164) is now  
2-2.4 times faster.  We have been able to get very reasonable  
pricing from DEC.  I have no firsthand info about running Linux  
though, we run Digital UNIX.

I am glad to provide benchmarks if anyone is interested for  
BLAST/FASTA for the following platforms.

2100 2X275MHz/256MB
4100 4X533MHz/1GB
ES40 4X500MHz/2GB

There are some other benchmarks at http://molbio.unmc.edu/Benchmark/ 


Chris