[Bioclusters] Workstation Selection for Bioinformatics Research

[Mark Komarinski]

On Thu, 2003-04-10 at 14:22, Andrew Fant wrote:
> > You need to be careful with your hardware selection if you expect to do
> > serious visualization work on your workstations. Make sure that whatever
> > graphics card / monitor combination you get is _well_ supported by the
> > X11/Linux distro you plan to use. It may be worth aquiring a test
> > machine before you actually commit dollars to a bigger purchase. It is
> > probably also worthwhile to try to find people who are currently using
> > any workstation combo you plan to aquire to see what real users think.
> 
> I'll go way out on a limb here and say that if you are going to want to do
> serious visualization under Linux, you will want a NVidia video card.  I
> know that free-software purists will object to their binary-only drivers,
> but I have been using them on this box for 6 months now without a hitch,
> and it has run much faster than anything that uses DRI (like the Radeon),
> and the glx support has been far more robust than a radeon or with the
> intel chipsets.  Not the cheapest solution, but one that you probably
> won't regret either.

If you want to do Stereo visualization, NVidia is the way to go.
The Quadro 750s can be had for under $400 and include the 3-pin DIN
for connecting standard stereo emitters.  We've recently started buying
goggles/emitters from NuVision for about 1/3 the cost of Stereographics
gear.  So far, they're working fine.

BTW, stereo under anything other than NVidia requires that you buy Xig's
Accelerated X.

-Mark

-- 
Mark Komarinski				mkomarinski@hms.harvard.edu
Sr. Linux/UNIX System Administrator	http://wqcg.med.harvard.edu
West Quad Computing Group
Harvard Medical School