Page 2 of 2 FirstFirst 12
Results 11 to 14 of 14

Thread: HPL-GPU 1.0 Released For GPU Super Computing

  1. #11
    Join Date
    May 2009
    Posts
    16

    Default

    Quote Originally Posted by deanjo View Post
    Too bad he didn't mention which Tesla card he was using for comparison.
    I was referring to the approximate number NVIDIA publishes for the Tesla devices. I don't have any current Teslas, so I cannot tell exact numbers.

    Quote Originally Posted by madjr View Post
    i wonder why they didnt add support for the open driver :O
    The open driver currently neither supports CAL nor OpenCL, so it's (nearly) impossible to use it for GPGPU.

  2. #12
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,582

    Default

    Quote Originally Posted by Marix View Post
    I was referring to the approximate number NVIDIA publishes for the Tesla devices. I don't have any current Teslas, so I cannot tell exact numbers.
    So you are unable to qualify the "The code also stomps NVIDIA, reaching 497 flops DGEMM on the HD5870 where NVIDIA only reaches around 300 on the much more expensive Tesla systems" statement. That is a bit misleading don't you think as you have no actual verification?

  3. #13
    Join Date
    May 2009
    Posts
    16

    Default

    NVIDIAs marketing number is in that area. I don't think they would understate their performance.

  4. #14
    Join Date
    May 2009
    Posts
    16

    Default

    You can get the NVIDIA number from slide 12 of one of NVIDIAs presentations from SC10: http://www.nvidia.com/content/PDF/sc...llips_SC10.pdf

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •