Re: [ccp4bb]: Xfit hardware stereo

["Harry M. Greenblatt" <harry.greenblatt@weizmann.ac.il>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

BS"D

Dear CCP4 users,


> Consequently the line quality in the vertical direction suffers a bit but that
> must be the same on the SGI. 

  Not on the SGI's that have been coming out for the past couple of
years.  They use quad buffered stereo which causes no loss of vertical
resolution.  The video memory holds 4 image buffers: left (front and
back) and right (front and back).  Only a few third-party cards can do
this:  3Dlabs Oxygen GVX1 and higher, and the Wildcat cards.  But linux
drivers from XIG graphics (www.xig.com) are so far available only for
the Oxygen family; Wildcat support is "coming soon"
...  I've heard that the Nvidia Quadro2 Pro can do quad buffered stereo,
marketed by Elsa as the Gloria III, but their web site doesn't seem to
say that the GloriaIII can do this, or that it has the serial port for
the stereo emitter.  Perhaps others could comment.

  But according to the following real application test I did, if stereo
hardware is not an issue, then you really don't need SGI at all.  Our
system administrator bought a simple PC PIII with a cheap Nvidia geforce
2 card for graphics.  We were interested to see how this performed
against our 3 year old Octanes (ESI graphics).  So we installed the
Nvidia linux drivers.  I took acetylcholinesterase (535 aa) with an
inhibitor bound and generated the surface of the protein with MSMS. 
Then I displayed this in DINO, contouring the surface by distance from
the inhibitor in increasing steps of 5A radius.  At first I had maybe 8K
triangles.  At the very end I had 270K triangles (whole surface of the
monomer).  The octanes became noticeably slower at ~140K triangles, and
were definately jerky at 270K (but not impossible).  The simple PC (at
the same screen resolution), was only getting a bit slow at 180K, and
was only a little bit jerky at 270K triangles.  I showed this to our SGI
reps here, and they said:  "Well you really should try the latest Octane
graphics, the V12."  So they loaned me an R12K Octane with V12
graphics.   At 270K triangles the V12 was definately slower, but not
quite jerky.  So it was a bit better than the Geforce2, but not 10x
better (which is the price difference).  But consider that the Geforce3
and Quadro2 Pro are even faster, and not much more expensive, they are
more attractive than SGI.  


Harry

-- 
------------------------------------------------------------------------
  Harry M. Greenblatt           | Phone: 972-8-934-3625
  Research Assistant            | FAX  :           4159
  Dept. of Structural Biology   |
  Weizmann Institute of Science |
  Rehovot, Israel 76100         | Email: harry.greenblatt@weizmann.ac.il
------------------------------------------------------------------------