VMD-L Mailing List
From: John Stone (johns_at_ks.uiuc.edu)
Date: Sat Mar 16 2019 - 01:11:39 CDT
- Next message: Tom Bartl: "32bit memory problems on Mac OS"
- Previous message: John Stone: "Re: ORCA Orbital Bug"
- In reply to: Casalino, Lorenzo: "Re: RTX-accelerated VMD 1.9.4 test build for 64-bit Linux posted..."
- Next in thread: John Stone: "Updated RTX-accelerated VMD 1.9.4 test build for 64-bit Linux posted..."
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Lorenzo,
Sorry for the slow reply, I've been traveling almost nonstop for
a couple of weeks, and I'm still far behind on email as a result.
My answers are inline below:
On Tue, Feb 26, 2019 at 07:02:03PM +0000, Casalino, Lorenzo wrote:
> Hello,
> Thank you John for this information.
> I have a few questions regarding this important update:
> 1) Which NVIDIA GPU RTX models would you recommend to better (and fully)
> exploit the performance improvements offered by this update?
The Quadro RTX 6000 is probably the best GPU for RTX due to both its
high compute performance and large GPU memory capacity. It's an expensive
card, but it's the ideal choice for visualizing huge structures.
If that card is too costly, then the next option would be the
new RTX Titan GPUs, which have a good amount of memory and are fast
also. They don't have all the same features as the Quadro, but
they should be a great choice also. We've ordered a bunch of the
RTX Titans, but they haven't arrived yet, so my own personal experience
is currently limited to the GeForce RTX 2080 (only 8GB RAM) and the
Quadro RTX 5000 and 6000 GPUs.
> 2) Dees this VMD update improve a little the TachyonLOPtiX ray tracing
> performance even if different NVIDIA GPUs are used (like Quadro P6000)? Or
> is it strictly specific for RTX models?
There will likely be a small improvement on the P6000, but the
new RTX "Turing" GPUs are something on the order of 6x to 10x faster
than the Quadro P6000, so really if you want the speed, you'll
need to get a new RTX GPU.
> 3) Which kind of benefits should I expect when dealing with 100 million
> atoms systems in terms of loading a state, interacting real time with the
> system, changing representations, visualizing a trajectory and
Loading state files will get better from a different revision I'm in
the middle of making to VMD. It will get 10x to 20x faster than it is
presently when I have the changes finished. This is unrelated to the
new GPUs however, and I'll send you a note when you can test it fully,
hopefully in the next week or two, as 80% of the code is already implemented.
The RTX GPUs have only minor impact on anything but the ray tracing performance.
The other uses within VMD will only be maybe 1.2x to 1.5x faster than what
you've seen on a P6000.
> rendering images/movie?
Rendering w/ the ray tracing engines is where the speed is obvious...
Best regards,
John Stone
johns_at_ks.uiuc.edu
> Thank you
> Best
> Lorenzo
> - - - - - - - - - - - - -
> Lorenzo Casalino, PhD
> Post-doctoral research scholar - Amaro Lab
> University of Califonia, San Diego
> 3221 Urey Hall, La Jolla, CA 92039-0340
> [1]https://amarolab.ucsd.edu/
>
> On Feb 25, 2019, at 6:40 PM, John Stone <[2]johns_at_ks.uiuc.edu> wrote:
> Hi,
> Now that the NVIDIA drivers with RTX support are generally
> available, I have made a current 64-bit Linux test build of VMD 1.9.4
> (a29)
> that includes RTX hardware-accelerated ray tracing, and it is
> posted on the web site. To use this version you will need the latest
> NVIDIA 418.xx (where xx >= 30) driver version from the NVIDIA web site.
>
> VMD users should expect TachyonLOPtiX ray tracing performance gains
> that range as high as 8x faster when comparing the new RTX GPUs with
> the new VMD with prior versions of VMD running on so-called
> Pascal and Volta class NVIDIA GPUs.
>
> Best regards,
> John Stone
> [3]vmd_at_ks.uiuc.edu
>
> --
> NIH Center for Macromolecular Modeling and Bioinformatics
> Beckman Institute for Advanced Science and Technology
> University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
> [4]http://www.ks.uiuc.edu/~johns/ Phone: 217-244-3349
> [5]http://www.ks.uiuc.edu/Research/vmd/
>
> References
>
> Visible links
> 1. https://amarolab.ucsd.edu/
> 2. mailto:johns_at_ks.uiuc.edu
> 3. mailto:vmd_at_ks.uiuc.edu
> 4. http://www.ks.uiuc.edu/~johns/
> 5. http://www.ks.uiuc.edu/Research/vmd/
-- NIH Center for Macromolecular Modeling and Bioinformatics Beckman Institute for Advanced Science and Technology University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801 http://www.ks.uiuc.edu/~johns/ Phone: 217-244-3349 http://www.ks.uiuc.edu/Research/vmd/
- Next message: Tom Bartl: "32bit memory problems on Mac OS"
- Previous message: John Stone: "Re: ORCA Orbital Bug"
- In reply to: Casalino, Lorenzo: "Re: RTX-accelerated VMD 1.9.4 test build for 64-bit Linux posted..."
- Next in thread: John Stone: "Updated RTX-accelerated VMD 1.9.4 test build for 64-bit Linux posted..."
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]