OTOY Forums

p3taoctane wrote:I have a Quadra 3700 and GTX480 in a PC i7 with 8 megs of Ram
Running the latest demo version 1.022 Beta 2.2
Installed Cuda toolkit 3.0 also

I run EVGA Precision to see how the GPUs are working and only GPU 2 is active.

Is that a Demo limitation?

If not... Is there anything within the Octane "software" that can fix this. I am talking with Cubix about getting an Octane Cubix render to add GPUs to the mix to increase my VRAM as I generally work with large OBJs (A gigabyte and up) and having tired to import one already it said too complex. I am hoping thats because it is only seeing one card. If I am able to have Octane see all the cards will that improve that situation? Is there a practical limit for GPU rendering

Many Thanks

It doesn't automatically use all GPU. It's under device manager inside Octane. You have to add both cards in and then Octane will use all of them. Also, I think you are having some technical misunderstanding of how VRAM works. VRAM is local memory for GPU. GPU can't access the VRAM on different cards. When you do a render with Octane, all cards will have duplicate info in VRAM. So if you are using multiple GPU, the amount of RAM on your smallest VRAM Video card will be the amount you can utilize. Getting Cubix GPU-Xpander doesn't do squat of helping you expand your VRAM. Only way to do that is to buy a Quadro 5800 (seeing you have a GTX480 already so that's 1.5GB and Quadro 5800 is 4GB of DDR3 RAM). But at the same time, you can't use that max amount of RAM unless you only do single GPU render on that Quadro 5800.

Peter

p3taoctane wrote:thank you very helpful. it seems the best way to go for heavy poly scenes then is to get the card with the most ram. Like you suggested a Quadra 5800.

not like I have money to burn... but is a tesla even better then or is that something altogether different

Peter

p3taoctane wrote:I have a Quadra 3700 and GTX480 in a PC i7 with 8 megs of Ram
Running the latest demo version 1.022 Beta 2.2
Installed Cuda toolkit 3.0 also

I run EVGA Precision to see how the GPUs are working and only GPU 2 is active.

Is that a Demo limitation?

If not... Is there anything within the Octane "software" that can fix this. I am talking with Cubix about getting an Octane Cubix render to add GPUs to the mix to increase my VRAM as I generally work with large OBJs (A gigabyte and up) and having tired to import one already it said too complex. I am hoping thats because it is only seeing one card. If I am able to have Octane see all the cards will that improve that situation? Is there a practical limit for GPU rendering

Many Thanks

Peter

radiance wrote:

p3taoctane wrote:I have a Quadra 3700 and GTX480 in a PC i7 with 8 megs of Ram
Running the latest demo version 1.022 Beta 2.2
Installed Cuda toolkit 3.0 also

I run EVGA Precision to see how the GPUs are working and only GPU 2 is active.

Is that a Demo limitation?

If not... Is there anything within the Octane "software" that can fix this. I am talking with Cubix about getting an Octane Cubix render to add GPUs to the mix to increase my VRAM as I generally work with large OBJs (A gigabyte and up) and having tired to import one already it said too complex. I am hoping thats because it is only seeing one card. If I am able to have Octane see all the cards will that improve that situation? Is there a practical limit for GPU rendering

Many Thanks

Peter

I don't think you can mix geforce and quadro's in one machine for octane.
you can use a quadro for display of a render rendered on a 2nd geforce, but you must install both drivers, and connect your display to the quadro, and leave the Geforce headless (no monitor)

Radiance