Preprocessing

Preprocessing

GPU rendering problems

    • rodriguez
      Subscriber

      Hello,


      I am working with pretty big geometries/meshes in both SpaceClaim and in Fluent mesher standalone and are having trouble working fluently.


      We have a GPU of 4 Gb dedicated memory and when it gets full, the program stops responding. Meanwhile the shared memory stays in zero.


      Do you know if there is any setting or something that must be modified in order to properly work?


      Thanks in advance.

    • peteroznewman
      Subscriber

      Bigger models need bigger hardware capabilities. What is the make and model of the display adapter?

    • rodriguez
      Subscriber

      Hi,

      Yes, I understand that.

      I guess I will have to buy one. Can anybody suggest me what I could buy? I mean. The GPU will not be used for simulating, "only" to render results.

      Is the dedicated memory the only parameter I should care about? The GPU I currently have is a Nvidia Quadro, I don't really know if I could also use a GeForce, for example. RTX I understand is also an option, being also for professional use.

      If the memory is the only important thing, I could also get a cheap Intel Arc...

      Any input would be appreciated.

      Also regarding the Nvidia control pannel settings. Any idea about what is the best performance settings for SpaceClaim and Fluent Standalone??

      Thank you very much in advance. I don't have much knowledge about GPU world at all...

      Kind regards.

    • peteroznewman
      Subscriber

      Ansys divides GPUs into two categories, one for display and one for accelerating computation.  Ansys tests a limited number of GPUs and lists the ones tested in a document.

      Here is the pdf file that has the display adapters:

      https://www.ansys.com/content/dam/it-solutions/platform-support/2022-r2-graphical-display-graphics-cards-tested.pdf

      While a GeForce card will probably work for the display, it hasn’t been tested. In a professional environment, cards from the published list are recommended.

      Here is the pdf file that has the list of GPU accelerator cards

      https://www.ansys.com/content/dam/it-solutions/platform-support/previous-releases/ansys-2022-r2-gpu-accelerator-capabilities.pdf

      In the past, these two lists had separate cards, but now I see the Nvidia RTX A4000 card is on both the display list and the accelerator list for Fluent!  This card has 16GB of GPU memory, so that might be a better choice for your needs than the RTX 4000 which has 8GB of GPU memory.

    • rodriguez
      Subscriber

      Thank you very much. Very useful information.

Viewing 4 reply threads
  • You must be logged in to reply to this topic.