[extropy-chat] visualizing nanoscale

[Eugen Leitl]

German transhumanists were discussing PR issues, when the
question of decent (c)-free images came up. Specifically,
we were talking about realistic images of nanomed gadgetry.

A suggestion was to roll your own. A couple of days ago
I came upon this nice semilog-step flash visualization:

	http://cellsalive.com/howbig.htm

I stole it, and attempted to add a couple of steps, to
bring it down to the nanoscale. After two hours (and
lots of growling) I got this:

	http://moleculardevices.org/howbig.htm

The first item above is the flash I shamelessly stole from
the fine folks at cellsalive.com. (Yay! They so rock. Kudos.
It's just for the sake of illustration, and hence arguably
fair use. I'll take it down soon. That should be sufficient grovelling).

The final step is a rhinovirus; I couldn't find one in a 
hurry, so I took a simian virus structure (SV40 capsid, a
66 MByte ~pdb file). Next to it we see the usual suspects:
neon pump, a wet lipid bilayer (crystallized, it's really cold 
in there), a planetary gear, and the fine motion controller.
All lifted from IMM, of course. I notice we haven't had 
new designs in a long while. It's really just up to a couple
of key people. They do nothing, nothing happens. This is not good.

The next step below is a closeup, minus the virus particle.
This is full atomic resolution. 

This was done on somewhat vintage hardware (Mac G4, 1.5 GByte RAM,
oldish OpenGL accelerator). I'll try this again this weekend
on a more recent (but: not bleeding edge system). VMD and PyMOL
were both being able to handle the image size. I haven't tried
Jmol, but have a suspicion it might be up to the task. The weak
point was, surprisingly, Povray. The raytracer could barely
render a 132 MByte .pov file with 1.5 GBytes RAM. Perhaps somebody
here can suggest a better deal? Tachyon? Something else?

All in all the handling of structures was very sluggish. 
I suspect a current machine (G5/Opteron, Radeon 9800) would
allow me to handle systems of this size almost in realtime.
This is just trivial manipulation, structure minimization
or MD are completely out of question here. I'll try VMD/NAMD
later on, to report which system size can be currently treated
in realtime on more or less recent (not bleeding edge) hardware.

All in all, our software is pathetic. So is our hardware. 
We need both lots of memory, lots of crunch over that memory,
and good memory bandwidth (OpenGL doesn't scale here). Here's
definitely a case of a system which would profit from a desktop
Blue Gene (which reminds me: <http://moleculardevices.org/bluematter.pdf> ) 

-- Eugen* Leitl <a href="http://leitl.org">leitl</a>
______________________________________________________________
ICBM: 48.07078, 11.61144            http://www.leitl.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE
http://moleculardevices.org         http://nanomachines.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20031121/6de3c464/attachment.bin>