Using old NVidia Cards in DAZ

Hello all, I just made a great discovery Rendering with Iray in DAZ.   Last year NVidia stopped supporting Kepler Chipset GPUs and Covid had more or less wiped out stock of new NVidia Geforce Graphic cards. Until then I was using  an old GTX 670 that was ok until DAZ v4.1 was updated, this was after the last Dirvers for Kepler GPUs had been released. I found the new Iray render engine stopped working in GPU mode so I did a bit of hunting on the internet for second hand cards that would work, being other than Kepler Chip set. There were newish GeForce Card on eBay but the bidding was mad. So I then looked at Quadro Cards and found the Kx200 range was Maxwell Chip set. I managed to get a 4Gb NVidia Quadro K2200 and it worked well not much slower than my old GeForce GTX 670 but it had more Ram. I then  gambled on using 2x Quadro K2200 cards and it has worked better than I expected. The Iray GPU Render time is ½ of using 1 card, I did not expect it to work at all as I saw some comments somewhere that only 1 card would register for use in Windows 11? As the Quadro K2200 is not SLU compatible.  As I only have 1 PCIe x16 slot on my Motherboard so I used a PCIe x1 to x16 Riser card. I expected the second card to be slower as in PCIe x1 slot but My Motherboard has PCIe Gen 3,  I think it helps with the Quadro K2200 being PCIe 2 x16 slower than my PCIe Gen 3. OK things are not supper fast but I did it on a low budget. I got the Quadro K2200 cards from eBay £40 each much cheaper than a new GeForce so I am happy for now, until NVidia end support for the Maxwell Chipset. Things in Daz seem to run a lot smother with faster refresh rates etc. But befor you go out hunting for cheap Quadro’s check the chipset spec as most of the old Quadro’s are Kepler chipset.

So looking for a cheap Daz Iray Render PC a second hand Quadro is the answer and the Quadro K2200 runs on 68W so will not need a big expensive PSU.  

Comments

  • The 2 Quadro K2200 Cards are showing up in Render settings Tab as CUDA 0 and CUDA 1 That is great as did not expect it to work !

    Untitled.jpg
    1920 x 1080 - 337K
  • I thought Kepler, which is the K in your cards, was no longer supported. Its days are certainly numberd so make sure you back up your isntallers for DS and its plug-ins and be careful to check compatibility for updates.

  • Richard Haseltine said:

    I thought Kepler, which is the K in your cards, was no longer supported. Its days are certainly numberd so make sure you back up your isntallers for DS and its plug-ins and be careful to check compatibility for updates.

    Be aware that nvidia uses the same prefix for several different 'pro' cards, so it's basically meaningless to determine which series it is.

    K2200's are gm107(GTX 750's).

    M2090's are fermi but M40's are maxwell.

     

     

  • no idea of current pricing but I still use a titanx 12g I got on ebay 4 years ago for $400...
    Outside of out and out speed the 3060 12 doesn't improve the cuda count 
    So for a real upgrade I need to go to the 80/90s 
    but theoretically if you use the quadro cards the will pair with the 24k teslas (because the studio driver will see the teslas. 

  • alan bard newcomer said:

    no idea of current pricing but I still use a titanx 12g I got on ebay 4 years ago for $400...
    Outside of out and out speed the 3060 12 doesn't improve the cuda count 

    But because of the much more efficient architecture and the additional Tensor and RT cores, the 3060 beats a Titan X easily in 3D rendering.

  • stefan.hums said:

    alan bard newcomer said:

    no idea of current pricing but I still use a titanx 12g I got on ebay 4 years ago for $400...
    Outside of out and out speed the 3060 12 doesn't improve the cuda count 

    But because of the much more efficient architecture and the additional Tensor and RT cores, the 3060 beats a Titan X easily in 3D rendering.
     

    I figured it would be faster but since I render in one instance while I do other work ... speed isn't the issue as much as getting the extra ram. 

     

  • PerttiAPerttiA Posts: 10,024

    alan bard newcomer said:

    stefan.hums said:

    alan bard newcomer said:

    no idea of current pricing but I still use a titanx 12g I got on ebay 4 years ago for $400...
    Outside of out and out speed the 3060 12 doesn't improve the cuda count 

    But because of the much more efficient architecture and the additional Tensor and RT cores, the 3060 beats a Titan X easily in 3D rendering.
     

    I figured it would be faster but since I render in one instance while I do other work ... speed isn't the issue as much as getting the extra ram. 

    What extra RAM? 

Sign In or Register to comment.