OTOY Forums

I have a gtx 580 (3GB) and a Titan (6GB) in my system. For some reason it seems to be still limiting me to 3GB Ram even when I disable the GTX 580 in my devices panel. As such I still cannot load my scene into octane on the titan, render fails if I exceed 3GB in all cases.

darkline wrote:I have a gtx 580 (3GB) and a Titan (6GB) in my system. For some reason it seems to be still limiting me to 3GB Ram even when I disable the GTX 580 in my devices panel. As such I still cannot load my scene into octane on the titan, render fails if I exceed 3GB in all cases.

We have many gpu's here 780 (6GB) all register their VRAM correctly. We have one titan 6GB, it always registers hardly any vram. barely 3GB. On a fresh restart of windows, it registers ~4/5 GB-ish. but not long after it is vram capacity seems to vanish.

Any illumination of the problem would be helpful, as the titans are expensive and currently not using their capability.

@darkline, a workaround for rendering is to send a backburner render to your own machine so that you dont have the max scene open at the same time. This frees up vram for rendering.

@Karba
Can you please consider the ability to revert back a proxy to poly object (as vray) because sometimes it is hard to find original object to mofify it... Thanks

Hi Karba,

thanks for new release. Is it possible to fix bug whit timeouts in 2.06 or we must wait for new release? Version 2. is very unstable for big scenes and I have a lot of problems from beginning. In this version you fix memory bug, great-thanks! But there is another big bug whith timeouts.
If you are able fix this bug as soon as possible I will be happy.

Thanks a lot
Jan

hi karba
we have the following problem with the 2x versions (on two different systems, both max design 2012 x64):
max crashes when closing a scene and ask if we want to save a copy. the scenes are NOT corrupt afterwards.
- the crash happens only on some scenes (bigger scenes)
- could be the crash happens only on scenes which were edited with pre-2x versions of octane
- the crash won't happen if we just edit the scene without ever opening octane preview window

i hoped maybe this was fixed with the memory bug but it wasn't.

it's minor but a bit annoying..

jan kudelasek wrote:Hi Karba,

thanks for new release. Is it possible to fix bug whit timeouts in 2.06 or we must wait for new release? Version 2. is very unstable for big scenes and I have a lot of problems from beginning. In this version you fix memory bug, great-thanks! But there is another big bug whith timeouts.
If you are able fix this bug as soon as possible I will be happy.

Thanks a lot
Jan

That should be fixed in 2.06, I just forgot to mention. Did you try 2.06?

yes, same problem in 2.06.

jan kudelasek wrote:yes, same problem in 2.06.

How many samples per second do you have?

Perhaps I'm not being clear.

The titan Vram clearly says 5.(something) GB avaiable in the octane window, but render fails everytime my Vram exceeds 3GB. The only thing I can think of is that I have a 3GB 580 in the system too (even though it's disabled in octane device panel). This did not occur in Octane 1.x

Karba wrote:

jan kudelasek wrote:yes, same problem in 2.06.

How many samples per second do you have?

samples/s are about 0,5-1,2
maybe I was wrong, because when rendering failed, it looked like this problem, the same message from windows, but I found the error log is different. However, with version 2.x I have similar problems with large scenes. I have problems with scenes rendered on around 3-4 gigabytes bigger resolution 3.5k-5k. I use forestpack, displacement. I noticed that one scene only failed with displacement on the carpet. Failure usually happens in every scene in the same time. Always I have the same message - windows disconnected from the nvidia driver.Look at the attachment please.

Started logging on 28.08.14 15:56:39

OctaneRender version 2.06 (2060000)

CUDA error 999 on device 0: An unknown internal error occurred.
-> kernel execution failed (pt)
device 0: path tracing failed