SUN vs SGI

Justin Diachun diachun at ncifcrf.gov
Tue Apr 27 08:51:00 EST 1993


In article <1993Apr27.001246.17373 at kakwa.ucs.ualberta.ca> chris_upton at darwin.biochem.ualberta.ca writes:
> I'd like to know a little about the groups ideas on preferences between
>SUN and SGI workstations.

I help administer both kinds of workstations here.

>At the moment I use a Sparc 2 to run the GCG
>package and also the Staden package (for ABI sequence data) with a few
>other programs such
>as MASE and PRIMER.
>I know GCG will run on SGI, does anyone have the Staden package running
>on SGI? Output is to X-windows but is this SUN specific? 

We use Staden too, and I was wondering this myself.  None of the X
stuff seems Sun specific from what I see.  I considered trying to
compile it on an SGI but it seems like it would be a major project to
get it working with confidence.  I asked around once and nobody
claimed to have done it.  Someone please speak up if they have; I
would prefer to switch over to our SGI system where we have more disk
space and free cpu cycles, plus this is where we have our (unix) GCG
package running.

>There's also a great program from Michael Zuker for sub-optimal dot plots
>on the SGI only.

Maybe I have been sleeping.  Can I get this?  Where?  How?

>But other considerations....?  Cost for comparable machines (I'm looking
>into it),

Well, the SGI machines do tend to be more expensive.  You get some
excellent graphics capabilities for the deal (if you use them).

> ease of administration (I'm no programmer).

I would pretty much call them even here.  I tend to think the SGI
hardware doesn't cause many problems and is easier to learn and
restart when something goes wrong.  But not by very much.  One big
advantage for Sun is that many new software packages that you try to
build will go up much easier on a Sun, because that is most likely the
platform they were developed on.  And some are in developing-only
stages or only ported to one platform (hmm--like Staden).  So you
might have more potential "new" programs to install and toy with if
you like, on a Sun.

>Probably better
>graphics packages for displaying protein structures on the SGI, this could
>be important.

Most of our other groups working on SGI equipment are heavily into
protein modeling and similar stuff.  They mostly rely on the
commercial packages like Insight, Quanta, and Sybyl that are really
only available for SGI workstations.  If you never plan to buy any of
these or do any 3-D modeling, then that takes away a major edge that I
see in having SGI machines.  A lot of our scientists that use these
SGI modeling packages hardly touch GCG or Staden type software, so
they know which machine they want...

>Are the SGIs moving away from being "crystallographer-only machines"? 

Certainly, yes!  Almost everyone seems interested in them these days.

>Thanks for any advice,

If you have a big enough group to afford a dozen or more machines, it
might make sense to buy at least 1 SGI machine to see if you like what
you get.  Although the headaches of administering a single "different"
machine might not be worth it and it may sit unused.  Maybe try to
convince the Chemistry department to buy one and have them show it off
to you sometime?  Just a thought.

>Chris Upton
>Biochemistry
>University of Alberta


Justin Diachun
National Cancer Institute
Frederick Biomedical Supercomputing Center


-- 
___   ____   ____   ____   ___________________   ____   ____   ____   _
\  \ / \  \ / \  \ / \  \ /      Helix        \ / \  \ / \  \ / \  \ / 
/\  \  /\  \  /\  \  /\      Justin Diachun    \  /\  \  /\  \  /\  \  
  \__\/  \__\/  \__\/  \   diachun at ncifcrf.gov  \/  \__\/  \__\/  \__\/




More information about the Bio-soft mailing list