1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

Question about cuda and kernels and shaders and such.

Discussion in 'Nvidia' started by Skybuck Flying, Dec 10, 2007.

  1. Hello,

    Cuda and the new nvidia cards seem to work with kernels and such which can
    execute code.

    I wonder about these kernels and the instructions they execute.

    Do all instructions take the same ammount of cycles or is there a difference
    ?

    If they are different what methods exist to keep kernels in sync ?

    What about the older model... like shaders.

    Do shader instructions execute in sync ?

    Bye,
    Skybuck.
     
    Skybuck Flying, Dec 10, 2007
    #1
    1. Advertisements

  2. Well from reading the cuda programming guide, it seems it's a bit messy,
    anything can execute at different speeds, order and time and there are some
    synchronization api's or synchronization objects and stuff like that. There
    are like kernels, threads, thread blocks, warps, some atomic functions,
    doesn't seem to have any special parallel instructions or funtions.. seems
    like a single program is applied to multiple data... I think they call it
    "mapping" of data or so... all very complex it seems.

    Don't know about shaders though.

    Bye,
    Skybuck.
     
    Skybuck Flying, Dec 10, 2007
    #2
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.