Compilation problem with CUDA, CMake, VS2012 on Windows

Front page Forums Installation Compilation problem with CUDA, CMake, VS2012 on Windows

This topic contains 17 replies, has 4 voices, and was last updated by  Dimitar 3 years, 8 months ago.

Viewing 3 posts - 16 through 18 (of 18 total)
  • Author
    Posts
  • #716

    Dimitar
    Member

    Hi,

    On your first question – whenever you call the .Solve() function the system will be solved in parallel (OpenMP, CUDA, etc). So it makes no sense to try to solve the three systems in parallel (it will not give you any performance benefits).

    We had tried this on CUDA 5.5 and with Linux, we will try to test it for Windows. I will keep you posted on that (but this might take couple of days).

    Best,
    Dimitar

    #717

    Wonder
    Participant

    Hi, Dimitar.
    But as I understand there is no way now to make one system be solved on CPU and another on accelerator IN PARALLEL. Not one system on CPU and THEN another system on GPU.
    The question is “Is Paralution thread safe library?”

    #719

    Dimitar
    Member

    Hi,

    Good question. We have never tried that – I think it should work. You need to have two fully independent problems (2 x matrix, rhs, x). And then in the first thread you move matrixA, rhsA and xA to the accelerator and in the second thread you can keep matrixB, rhsB, and xB on the host and solve them. This should work.

    Keep us posted on that.

    Best,
    Dimitar

Viewing 3 posts - 16 through 18 (of 18 total)

You must be logged in to reply to this topic.