Jump to content
Beckman Coulter User Forums

NVidia Tesla Card Compatibility


Recommended Posts

The Kaluza software optionally supports the Tesla line of GPU-based supercomputer boards from NVidia. NVidia frequently releases new versions of these boards and obsoletes older ones. This post explains which models the Kaluza software has been tested with.

 

C1060 - Formally tested with Kaluza v1.2

C2050 - Formally tested with Kaluza v1.2

C2070 - Informally tested, but based on silicon from the C2050 and we have high confidence in it

C2075 - Informally tested, but based on silicon from the C2050 and we have high confidence in it

K20 - Limited informal testing. It functions but we have not compared results

 

You can find out more about these products from the following websites:

http://www.nvidia.com/tesla

http://en.wikipedia.org/wiki/Nvidia_tesla

Link to post
Share on other sites
  • 7 years later...

Regarding supported Cuda GPU models.

In the Presentation from 2012:

https://on-demand.gputechconf.com/gtc/2012/presentations/S0638-Lenovo-Thinkstation-Accelerates-Medical-Research.pdf

You show different Lenovo Models and Nvidia in a PDF. Since then we do not find any information about supported cards. The GPU cards

mentioned in the actual manual are all EOL since a few years.

How about the "Nvidia Quadro RTX 4000 8GB" which HP ships with their Z2 and Z4 workstations with G4 Modells.

A little more info for such expensive software would be the least a customer could await. 

You can lend/rent GFX cards for such tests you don't have to buy them..

 

For CPU Workstation:

 

* Please recommend Intel CPU like i7 and i9 and with how much Cores?

* Please recommend if it would be useful to use XEON Workstation you just mention SSE2 from 200X?

* How much RAM | More RAM or higher CPU? How much percentage ratio CPU / RAM / GPU-RAM (Not decent give us a factor)

Thanx.

Greetings from Switzerland

Senior System Engineer IT

 

 

 

Link to post
Share on other sites
  • 2 weeks later...

I apologize for the delayed response.

You can use non-Tesla NVidia boards, but you have to pass the /CUDA command line switch to the software when it's started. Bear in mind that when you use a video board for computation that you're sharing the board's RAM. Kaluza won't use more than 4GB of RAM, so if your board has enough RAM for the display at your chosen settings and still has 4GB left over then there's no compromise. If there isn't 4GB of RAM left over, then the file size that Kaluza can use may be limited. You can use the About Box in the software to see if your CUDA board has been recognized.

It's difficult to recommend a specific CPU or RAM configuration as needs vary. Kaluza scales very well with CPU cores, having more CPUs and more cores per CPU will help performance. However, once the software is already fast enough for a given need, there's no incremental benefit to adding more cores. From a RAM standpoint, Kaluza is currently a 32-bit application and so cannot address more than 4GB of RAM. If there's enough RAM for the OS and other applications, leaving at least 4GB for Kaluza then that will be sufficient.

Note that if Kaluza is using a CUDA board for computations, then there's little incremental benefit to more workstation cores or RAM from a Kaluza perspective.

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...