The Notebook Review forums were hosted by TechTarget, who shut down them down on January 31, 2022. This static read-only archive was pulled by NBR forum users between January 20 and January 31, 2022, in an effort to make sure that the valuable technical information that had been posted on the forums is preserved. For current discussions, many NBR forum users moved over to NotebookTalk.net after the shutdown.

NVIDIA CUDA Tool Kit 7.5 / nvsmi-smi.exe

Discussion in 'Dell Latitude, Vostro, and Precision' started by ygohome, Mar 15, 2016.

Tags:
Thread Status:
Not open for further replies.
  1. ygohome

    ygohome Notebook Deity

    Reputations:
    210
    Messages:
    1,254
    Likes Received:
    158
    Trophy Points:
    81
    Hi,

    Any CUDA programmers here? I've been trying to teach myself to program with the CUDA Tool Kit. I'm in Windows and developing in Visual Studio 2013 using c++ and the CUDA libraries included in the toolkit along with some others (Thrust c++ libs). The toolkit is available for different Linux distros too. I've been enjoying it quite a bit. I probably sound ignorant if this is common knowledge but there is a very cool GPU reporting/diagnostic utility that comes with the NVIDIA drivers.

    NVIDIA System Management Interface (nvsmi-smi)

    C:\Program Files\NVIDIA Corporation\NVSMI\nvidia-smi.exe

    Documentation is in a PDF of that same folder but you can get the same info using -h option.

    My question is, have any of you CUDA developers tried switching your Quadro to use the TCC drivers? Tesla Computer Cluster. I understand that it brings a number of benefits when developing for CUDA and comes enabled by default for Tesla cards (as would be obvious). As I understand it, using nvsmi, I think I can enable TCC for my Quadro. The GPU would not be able to act as a display card, but instead solely as a compute GPU. I would then use the iGPU for displaying and use the dGPU for my Cuda development.

    Obviously I would want to be able to switch back and forth using/not-using TCC. I have the Quadro mostly for video/photo editing and would like to switch when no developing for CUDA. Is there anyone who has tried this.

    Thanks


    *notes from nvidia:

    The TCC driver mode provides a number of advantages for CUDA applications on GPUs that support this mode. For example:


      • TCC allows the use of CUDA with Windows Remote Desktop, which is not possible for WDDM devices.
      • TCC allows the use of CUDA from within processes running as Windows services, which is not possible for WDDM devices.
      • TCC reduces the latency of CUDA kernel launches.
    TCC is enabled by default on most recent NVIDIA Tesla GPUs. To check which driver mode is in use and/or to switch driver modes, use the nvidia-smi tool that is included with the NVIDIA Driver installation (see nvidia-smi -h for details).

    Note: Keep in mind that when TCC mode is enabled for a particular GPU, that GPU cannot be used as a display device.
    Note: NVIDIA GeForce GPUs do not support TCC mode.
     
    Last edited: Mar 16, 2016
Loading...
Thread Status:
Not open for further replies.

Share This Page