Re: [Closed REQ 5986]: Geomview on Dec Alpha

[daemon]

>My lab at MIT is purchasing a new workstation (probably a DEC Alpha 500). I am
>wondering how efficiently Geomview and Srface Evolver run on this platform, as
>compared to the SGI platform.  Do you have recommendations on the graphics
>accelerator that should be purchased along with the workstation (Dec offers a
>number of options). 

First, for evolver, probably its author Ken Brakke can say something --
write to brakke at geom.umn.edu.

As for geomview, I can't say anything definite -- we have no DEC Alphas at
hand to run performance tests on.  Your best bet would be to get one of the
graphics accelerators for which DEC supports their Open3D software package,
including Open GL libraries.  According to the OpenGL FAQ as of last March,
that's:

             PXG (all devices in the PXG family with z-buffers)
             ZLX-M1
             ZLX-M2
             ZLX-E1
             ZLX-E2

You'd want to get one of these which includes a Z-buffer.  The upcoming
release of Geomview supports Open GL, and has seemed to work fine on
the four Open GL implementations where we've tried it -- SGI's,
the software-only free Mesa library, Sun's new one for UltraSparcs
with Creator3D hardware, and one for a DEC Alpha at Univ. of Maryland,
where I think one group has a (no-longer-supported?) Kubota graphics card
with OpenGL libraries.

One modern price-performance point we do have: a Creator3D-equipped
Sun UltraSparc (170 MHz?) had geomview benchmarks comparable to a 175-MHz
SGI Indigo-2 with XZ graphics, at about 20-40 frames per second for scenes
with a couple thousand polygons.  I think both machines have list prices
in the $25-30K range.  SGI's with their new Impact graphics cost about
30% more, if I remember right (for the non-texture-hardware Solid Impact kind)
but geomview graphics are faster by about a factor of 5.

I don't know how DEC's graphics accelerators compare with any of these.  If
you'd like to try benchmarking, we can supply geomview source & benchmark code.

Hope this helps -- sorry it's so late in coming.

   Stuart Levy, Geometry Center