+ Reply to Thread
Page 1 of 2 1 2 LastLast
Results 1 to 10 of 11

Thread: How to take advantage of the Nvidia Maximus technology

  1. #1
    Administrator david.randle's Avatar
    Join Date
    Apr 2011
    Location
    San Diego, CA
    Posts
    1,252

    Lightbulb How to take advantage of the Nvidia Maximus technology

    Nvidia Maximus technology is a Quadro Driver feature that lets you choose which GPU's do CUDA processing and which card does the graphics processing.
    This has a huge advantage for people that tend to operate CAD/Photoshop/AfterEffect etc. at the same time as rendering as it lets one card (the Quadro) handle all graphics requests on the system and pushes the intense CUDA calculations to the Tesla card. This allows you to keep working and rendering on the same machine at the same time.

    To benefit from this technology you must be running on a Maximus system OR have at least two GPU's, one of which is a Fermi based Quadro card and the other of which should be a Fermi based Tesla card and the nvidia 275.xx or newer driver.

    Set up the following environment variable:

    Variable name: CUDA_VISIBLE_DEVICES
    Variable value: 0

    Setting this will leave the Quadro for graphics apps (for example our interface, photoshop, CAD software etc.) and it will only do CUDA based rendering on the Tesla card.

    If you want to use both cards, then set the variable value to 0, 1 and it will use both again.

    If you have a 4 GPU system the Variable value would read 0, 1, 2 if you want to leverage Maximums or it will read 0, 1, 2, 3 if you want CUDA to run on all cards.

    As a note, whenever you change this, you have to restart our program for the change to take effect.

    This is very new technology supported on the latest nvidia driver. We plan on building that ability to turn your cards on/off into our software in the future.

    Hope this helps,
    David Randle / General Manager / Bunkspeed

  2. #2
    Junior Member mpaz's Avatar
    Join Date
    Dec 2011
    Posts
    16
    Will this work on quadro fx5800 + Tesla c2050 or you need a quadro x000 in the equation?

  3. #3
    Administrator david.randle's Avatar
    Join Date
    Apr 2011
    Location
    San Diego, CA
    Posts
    1,252
    Since its a driver feature i think that will work with your configuration....Download the latest nvidia drivers and give it a try. If it doesn't, you can always just delete the environment variable. Its not going to harm anything.

    You will need a gpu monitor such as EVGA Precision (Just google it) to see which GPU's are being used after you set up the variable.
    David Randle / General Manager / Bunkspeed

  4. #4
    Junior Member mpaz's Avatar
    Join Date
    Dec 2011
    Posts
    16
    It works! But the variable value depends on your system. For example on my system the quadro sits on the second position starting from the bottom, so I set it up like so: 0,2,3
    The EVGA Precision is a great tool, thanks! The only thing is I tried to overclock the Teslas to max so it crashed, lol! I don't know if they are even meant to get overclocked.
    Another gpu monitor you can use is the windows gadget "GPU Observer" but it can't monitor the memory usage of the Teslas.

  5. #5
    Administrator david.randle's Avatar
    Join Date
    Apr 2011
    Location
    San Diego, CA
    Posts
    1,252
    Ok, good to know. Thanks for the update

    BTW: overclocking the GPU is a very risky thing to do....i would urge you to avoid that.
    David Randle / General Manager / Bunkspeed

  6. #6
    Junior Member Jimbo55's Avatar
    Join Date
    Feb 2012
    Posts
    2

    Dell/HP vs custom built machines with Maximus.

    Hey David,
    I had a question that is related to this. According to Nvidia, Maximus technology is not accessible without buying one of their authorized computers (Dell/HP/Lenovo/Fujitsu). But the thing that is confusing is that Maximus is not hardware its software, and as far as I can tell it is not separate from the drivers that go into the Nvidia cards. So my understanding of Maximus is that if you have a Quadro card (6000 lets say) and a Tesala 2070, plus the latest drivers, you have Maximus am I correct?

    -Jim

  7. #7
    Administrator david.randle's Avatar
    Join Date
    Apr 2011
    Location
    San Diego, CA
    Posts
    1,252
    Hi Jim -

    You are correct.

    With that said, nvidia promotes Maximus technology on the tier 1 oem manufactures, which is why you hear about HP, Dell and Lenovo offerings, however, it doesn't mean that it won't work on non OEM systems.

    Thanks
    David Randle / General Manager / Bunkspeed

  8. #8
    Junior Member Jimbo55's Avatar
    Join Date
    Feb 2012
    Posts
    2
    Thank you David!!!!!

    -Jim

  9. #9
    Junior Member Callizto's Avatar
    Join Date
    Dec 2011
    Posts
    8
    There will be a reason why this half of all posts were deleted?

  10. #10
    Administrator david.randle's Avatar
    Join Date
    Apr 2011
    Location
    San Diego, CA
    Posts
    1,252
    yes, i told you i was going to remove them because they had nothing to do with the thread. We have record of your support case in our tracking system. The forum was not the right place to communicate a solution to your issue.
    David Randle / General Manager / Bunkspeed

+ Reply to Thread

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