Results 1 to 9 of 9

Thread: Octane 2020 And Multiple RTX Video Cards

  1. #1

    Octane 2020 And Multiple RTX Video Cards

    I heard that Octane 2020 would be able to use the video RAM on multiple RTX cards, so that if you had 2 RTX 2080Ti cards, Octane would use 22 gig of video RAM, not just 11.

    Anybody know anything about this?

  2. #2
    Registered User
    Join Date
    May 2019
    Location
    cologne
    Posts
    86
    i have no octane. but I think that's normal what you're describing. I use the Iray Render from DAZ studio. So also a GPU render. And there the ram is added.
    i would be very surprised if that wasn't the case with octane

  3. #3
    Super Member OlaHaldor's Avatar
    Join Date
    May 2008
    Location
    Norway
    Posts
    1,008
    Where did you hear about this Paul?
    3D Generalist
    Threadripper 2920x, 128GB, RTX 2080 Ti, Windows 10

  4. #4
    Registered User safarifx's Avatar
    Join Date
    Mar 2005
    Location
    Berlin
    Posts
    712
    Quote Originally Posted by paulk View Post
    I heard that Octane 2020 would be able to use the video RAM on multiple RTX cards, so that if you had 2 RTX 2080Ti cards, Octane would use 22 gig of video RAM, not just 11.

    Anybody know anything about this?
    To alleviate the card’s texture count limitation, you can also set OctaneRender to use the host memory (RAM) through its Out-Of-Core settings, allowing you to use more textures and geometry than would fit in the graphic memory of the video card (VRAM). Since Out-Of-Core memory is shared between GPUs, you can not turn devices on or off while using Out-Of-Core features.

    However, OctaneRender 2018.1 and higher support NVIDIA NVLink, which allows sharing data between two GPUs via an NVLink Bridge, on supported RTX GPU configurations.

    https://home.otoy.com/render/octane-render/faqs/

    snip safx

  5. #5
    Registered User
    Join Date
    May 2012
    Location
    Virginia
    Posts
    530
    Quote Originally Posted by safarifx View Post
    To alleviate the card’s texture count limitation, you can also set OctaneRender to use the host memory (RAM) through its Out-Of-Core settings, allowing you to use more textures and geometry than would fit in the graphic memory of the video card (VRAM). Since Out-Of-Core memory is shared between GPUs, you can not turn devices on or off while using Out-Of-Core features.

    However, OctaneRender 2018.1 and higher support NVIDIA NVLink, which allows sharing data between two GPUs via an NVLink Bridge, on supported RTX GPU configurations.

    https://home.otoy.com/render/octane-render/faqs/

    snip safx
    This ^

    NVLink lets select nvidia RTX cards share not only their cores but also their VRAM.

    But you still have to have software that recognizes it, unless you have cards that allow you to enable Tesla Compute Cluster (TCC) mode.

    TCC essentially lets you "sleep" one card for display purposes and opt it out of the Windows driver architecture, so as far as the OS and other software will automatically assume you have a single card with both cards' number of cores and VRAM.

    Also, you can't connect a display to an RTX card in TCC mode.

    Unfortunately right now, only Quadro and Titan RTX cards can be put into TCC mode. All other RTX cards are still automatically put in WDDM mode.

    That may change in the future, though.

  6. #6
    Quote Originally Posted by RPSchmidt View Post
    This ^

    NVLink lets select nvidia RTX cards share not only their cores but also their VRAM.

    But you still have to have software that recognizes it, unless you have cards that allow you to enable Tesla Compute Cluster (TCC) mode.

    TCC essentially lets you "sleep" one card for display purposes and opt it out of the Windows driver architecture, so as far as the OS and other software will automatically assume you have a single card with both cards' number of cores and VRAM.

    Also, you can't connect a display to an RTX card in TCC mode.

    Unfortunately right now, only Quadro and Titan RTX cards can be put into TCC mode. All other RTX cards are still automatically put in WDDM mode.

    That may change in the future, though.
    So, AFAICT, NVLink would work with 2 RTX 2080 Ti cards, with maybe my old GTX 670 for display.

    As for "software that recognizes it", besides Octane, what would need to recognize NVLink? I'm not a gamer.

    Finally, it looks like the NVlink bridge requires the video cards to be placed in adjacent slots that are closer together than the slots on an AMD StrX4 mobo. Yes/No/Maybe?

  7. #7
    Registered User
    Join Date
    May 2012
    Location
    Virginia
    Posts
    530
    Quote Originally Posted by paulk View Post
    So, AFAICT, NVLink would work with 2 RTX 2080 Ti cards, with maybe my old GTX 670 for display.

    As for "software that recognizes it", besides Octane, what would need to recognize NVLink? I'm not a gamer.

    Finally, it looks like the NVlink bridge requires the video cards to be placed in adjacent slots that are closer together than the slots on an AMD StrX4 mobo. Yes/No/Maybe?
    Other software examples would be other 3d programs that use GPU rendering, data analytic software that uses the GPUs rather than CPUs, system emulation software for complex simulations, etc.

    The cards don't have to be in adjacent slots. The NVLink comes in two "sizes"; a two slot size and a four slot size. The four slot size would allow you to connect two cards where the first card is in the top PCIEx16 slot and the second card is in the bottom PCIEx16 slot.

  8. #8
    Quote Originally Posted by RPSchmidt View Post
    Other software examples would be other 3d programs that use GPU rendering, data analytic software that uses the GPUs rather than CPUs, system emulation software for complex simulations, etc.

    The cards don't have to be in adjacent slots. The NVLink comes in two "sizes"; a two slot size and a four slot size. The four slot size would allow you to connect two cards where the first card is in the top PCIEx16 slot and the second card is in the bottom PCIEx16 slot.
    Sounds like a plan!!

    It looks like the NVLinks on newegg, amazon, etc are card-specific -- i.e., you shouldn't/can't use an ASUS NVLink on GigaByte RTX cards. Or am I misinformed?

    Now, if only the RTX 3070 was out . . .

  9. #9
    TD/Animator lino.grandi's Avatar
    Join Date
    Jun 2003
    Location
    Rome
    Posts
    1,708
    Quote Originally Posted by paulk View Post
    I heard that Octane 2020 would be able to use the video RAM on multiple RTX cards, so that if you had 2 RTX 2080Ti cards, Octane would use 22 gig of video RAM, not just 11.

    Anybody know anything about this?
    Yes Paul, using NVLink you can take advantage of 22 GB of GPU ram if you're using 2 RTX 2080 Ti cards.

Bookmarks

Posting Permissions

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