VMD-L Mailing List

From: John Stone (johns_at_ks.uiuc.edu)
Date: Thu May 25 2000 - 22:14:34 CDT

Hi VMD-L,
  I thought I'd post another quick note about using VMD 1.5b1
with the XFree 4.0 and the DRI OpenGL capabilities therein.
Now that VMD 1.5b1 is up on the web site, you will see a new
binary distribution called "LINUX DRI OpenGL (Linux with DRI OpenGL)"
as one of the download choices. This is a dynamically linked build
of VMD, produced on a RedHat Linux box running XFree 4.0 with the
Mesa 3.2 RPM installed. In order to run this binary on your own
machine(s), they will probably have to have the Mesa 3.2 RPM
installed in order to get a libGLU.so.1 library onto your
system. If you've got another libGLU.so.1 already there, then
VMD ought to work ok with that.

Once you've got VMD installed on your machine, you can test to see
if you are getting hardware acceleration by comparing the speed of
the regular "Mesa" binaries of VMD against the speed of the new "DRI OpenGL"
binaries. (You can download both, and rename vmd_LINUX in your
install directory and replace it with the vmd_LINUX from the other one
if you want to see how much of a speed difference the new XFree 4.0
drivers are giving you)

Earlier tonight I tested a GeForce256-DDR board with the NVidia
beta drivers, and it performs somewhere around 20 times faster than
the regular Mesa version, depending on the complexity of the molecule
being displayed. (the difference is actually largest for small molecules,
because the Mesa version has an incredible amount of per-frame overhead
which is constant regardless of molecule complexity. When displaying large
molecule with thousands of atoms, the DRI OpenGL build of VMD is probably
only 5 or 10 times faster.)

So, we've now seen VMD run with XFree 4.0 and hardware acceleration on
the NVidia TNT2, NVidia GeForce256-DDR, and 3dfx Voodoo3. If you have
success running the new "DRI OpenGL" binaries of VMD with other cards,
we would really like to hear about it.

Hopefully someone will get a chance to try VMD on the new GeForce 2 GTS
boards on Windows and Linux shortly. This could be a particularly
impressive board because it has per-pixel lighting, which would make
VDW and other reps that use spheres look at LOT BETTER even at low
levels of sphere tessellation. This will probably require a little
tweaking of VMD's OpenGL startup code to take advantage of, but we're
anxious to see it in action.

Again, let us know what your experiences are with this new "DRI OpenGL"
binary we've made. We want the final released version of VMD 1.5 to
be the best yet, especially since Linux finally has good 3-D acceleration
support!

Thanks for your time,
  John Stone
  vmd_at_ks.uiuc.edu

On Fri, May 19, 2000 at 06:22:39PM -0500, John Stone wrote:
>
> Dear VMD-L,
> Since I know many of you are using, or know someone
> that uses VMD on Linux, I thought I'd spread some cheer..
> We've recently built pre-release versions of VMD 1.5
> on XFree 4.0, and gotten decent hardware accelerated
> OpenGL performance. So far we have tested it on a
> 3dfx Voodoo3, and an NVidia TNT2 board. For the release
> version of VMD 1.5, I intend to ship a pre-compiled binary
> that should work for people running XFree4.0, with whatever
> drivers they are using. The major difference between this
> binary and the regular Mesa build is that it will be dynamically
> linked against the OpenGL and X11 libraries.
>
> In theory, this new "DRI" build of VMD will "just work" for anyone
> running XFree 4.0 on Linux. The new configure script has a "DRI"
> build flag similar to the original "OpenGL" and "Mesa" flags, which
> is setup to work for XFree 4.0. I'm still working out the details of
> adding some simple code to allow VMD to detect when its getting
> hardware acceleration vs. when its using a software renderer on
> XFree 4.0. This will make it easier for people to determine if
> VMD is getting peak speed from their XFree 4.0 setup.
>
> I'm planning on releasing 1.5 beta 1 in the next few days, and I'll
> have one of these DRI builds available. I'm hoping that those of
> you running XFree 4.0 will give it a spin and send us mail letting
> us know how well it works for you.
>
> In reading the DRI / XFree 4.0 docs, they still do not support stereoscopic
> display modes on Linux, but this is something I assume will be coming
> along in future revs of DRI and XFree, its just a matter of time.
>
> So, I thought I'd send out this "heads up" message before I make the
> beta 1 binaries in a couple days. If anyone wants to, the code is
> already in the CVS tree, so feel free to try building for yourself
> if you're curious. Otherwise binaries will come in a couple days.
>
> You can get read-only VMD CVS access from:
> http://www.ks.uiuc.edu/Research/vmd/cvsrequest.html
> usage info is at:
> http://www.ks.uiuc.edu/Research/vmd/anoncvs.html
>
>
> John Stone
> vmd_at_ks.uiuc.edu
>
> --
>
> Theoretical Biophysics Group Email: johns_at_ks.uiuc.edu
> Beckman Institute http://www.ks.uiuc.edu/~johns/
> University of Illinois Phone: (217) 244-3349
> 405 N. Mathews Ave FAX: (217) 244-6078
> Urbana, IL 61801, USA Unix Is Good For You!!!
> 

-- 
Theoretical Biophysics Group   Email: johns_at_ks.uiuc.edu
Beckman Institute              http://www.ks.uiuc.edu/~johns/
University of Illinois         Phone:  (217) 244-3349
405 N. Mathews  Ave              FAX:  (217) 244-6078 
Urbana, IL 61801, USA          Unix Is Good For You!!!