Skip to content

GPU Parallel

New to LBM and OpenLB? Attend our upcoming one-week hands-on Spring School in Marseille/France from 19.05.2025 – 23.05.2025.

OpenLB – Open Source Lattice Boltzmann Code › Forums › on OpenLB › General Topics › GPU Parallel

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • October 23, 2023 at 5:00 am #7845
    Alexmine
    Participant

    Dear OpenLB developers,

    I would like to utilize the NVIDIA RTX40 series GPU for parallel computing. However, this series of graphics cards adopts a new architecture called Ada-Lovelace, and I am unaware of the CUDA-ARCH value for this new architecture.

    Sincerely,
    Alexmine.

    October 23, 2023 at 8:45 am #7847
    Adrian
    Keymaster

    The highest supported value for Lovelace is 89.

    FYI it is possible to use older target gencodes on newer GPUs, they will be transparently translated by CUDA.

    October 23, 2023 at 11:00 am #7848
    Alexmine
    Participant

    Dear Adrian,

    Thank you for your reply. I would like to know the CUDA-ARCH value of RTX4070, and whether RTX4070 can be used for parallel computing in OpenLB.

    Sincerely,
    Alexmine.

    October 23, 2023 at 11:02 am #7849
    Adrian
    Keymaster

    I already answered this question, please read my reply above.

    OpenLB supports CUDA-based GPU execution and as such supports any GPU that supports CUDA.

    October 24, 2023 at 2:11 am #7851
    Alexmine
    Participant

    Thank you for your response.

  • Author
    Posts
Viewing 5 posts - 1 through 5 (of 5 total)
  • You must be logged in to reply to this topic.