OTOY Forums

Octane is constantly crashing with my 2 GPUs. I've been running 2 GPUs for a while without a problem. With one of the recent Nvidia updates I started getting 'render failure' crashes as soon as I started moving around in my Cinema4D scene.

I updated to 4.05 R6 (as suggested in the forum) and took out one of the GPUs (GTX 980). No crashes. Was able to finish a 4 week project on one single 1080Ti without one crash. Rock solid.

Now I put in a 2080Ti (so now I'm running on one 1080Ti and one 2080Ti) and I'm at the point where I am getting the 'render failure' crash as soon as I move around in my scene. Tried the usual combinations of rolling back the driver, installing studio rather than game ready etc. No luck.

What am I doing wrong? Do I need a specific version or is there anything I can do to fix this? I'm on Windows10, Cinema4D R21.026, Octane 4.05 R6 and Nvidia Studio Driver 441.12 (same with previous game ready driver)

Again with just the 1080Ti, everything works fine, as soon as I put in a 2nd GPU, constantly crashing.

I'm not even expecting any RTX acceleration or anything. At this point I'd be happy if it just would not crash every time.

Any help would be much appreciated.

You actually said what's the solution. Downgrade to which driver was the stable for you.

I did roll back the driver in the device manager under properties. Now I am sitting on the game ready driver 436.48. Still crashing. Now it's grayed out. How do I go back even further?

Is there any documentation for this?

How is 'roll back the drivers until it works' a viable solution for something you're supposed to make a living with?

Why do you ask for documentation? It's a new issue with latest nvidia drivers. And not up to me as a plugin developer. It's not possible to do stable when you have issues that comes from drivers and there's different factors comes from hardware and driver combination. Because renderer is using the functionality from drivers. Make a clean install to one of Nvidia STUDIO driver.

Well..here is the thing: When I purchase a product and there seems to be a persistent problem with Nvidia drivers, I would expect that there would be at least some kind of effort to shed some light on that problem.

I don't know what's going on in the background. If you are not capable of (or unwilling) to solve that, maybe it's not even in your power to anything about it. All nice and good.

But how is "just roll back drivers" a viable response for an international company / brand / product? Certainly when people rely on that product to make a living?

I had to go back 5 (!!) versions of the Nvidia driver now to get it to work again.

Again, maybe it's not in your hands to do anything about it....but asking for documentation which drivers work in which scenario is really too much to ask for?? Wow, that's what I call customer service.

So basically when I want to keep using Octane for my client projects, I have to test every now and then. Maybe it works, maybe not. Updating drivers would pretty much be guesswork. Is that what you are saying?

You are blaming wrong person for issues that You have.

"I've been running 2 GPUs for a while without a problem. " - don't fix it if it's working - that's the most important thing to remember.

Drivers are something that allow Your components to talk with each other...

In over simplified world there are:
5 lines of CPU,
with 5x models each,
X 20 different models of motherboards,
x 10 different GPUs,
x 50 customized GPU designs..
(and in reality those numbers are much bigger)
Now multiply all that and think a bit.

It's nearly impossible to test each and every configuration, not to mention that You might have issues not with driver but with Your system: bios version, windows, other software, bad installation, overclocked hardware, heat issues - just to name the few.

It would be good to understand, that developers are doing their best and if they are pointing You obvious and straight forward solution, that You could use, please do that for Your own sake.

I totally get Tom's standpoint and also Aoktar's.
These are very complex systems we aremaking our living out of.
But i am in the same boat since some days and i don't know what to do. Plus it costs a lot of time and energy to test things out. I open up cinema and start a render. Nothing special, but with 3x1080ti it takes around 8 minutes. When i am in luck, the render finished and wrote the files to my folder. If not, cinema simply crashed without any warning or log file. If the render succeeded and i start a second one, c4d will crash with a 90% chance. If i remember correct i had to upgrade my driver to something like 431 to be able to work with redshift 3.0. But also with the driver before i wouldn't consider octane c4d as a stable plugin on my system, probably due to the nvidia drivers. I get render failures on a daily basis. I have to be very careful in ipr.
And now i work on a bigger project which has to be rendered soon and i am afraid that i have to sit in front of the computer all night to monitor the rendering.

An important note: the same c4d crashing i have with redshift.

I deinstalled EVGAs XOC monitoring tool. Nothing demanding in the background. Beefy PSU.

If anyone has an idea what i can do, please let me know.

System is Win10
3x EVGA 1080 Ti SC2
128 GB RAM
Octane 4.5 R7
C4D R20
denoiser on

Hi,
I have written you a Private Message.
ciao Beppe

glimpse wrote: "I've been running 2 GPUs for a while without a problem. " - don't fix it if it's working - that's the most important thing to remember.

I can't approve of this comment enough. "If it aint broke, don't fix it," is one of the Deep Truths of Production. There's a very good reason that production studios tend to stick with a stable pipeline for extended periods of time, letting others live on the bleeding edge. Getting finished work out the door on time and on budget is generally much more important than having the latest cool widget, or an incremental increase in render speed.

Yes, it can be terribly frustrating when things aren't working, but we have all chosen to work in a field that is constantly changing, on the cutting edge of technology. Our tools are dependent on literally thousands of variables being compatible with one another. 3D animation tools may someday be as stable and reliable as word processing or 2D image manipulation, but it's been in more or less the same state of semi-stable chaos for more than 30 years now, and I doubt that will change any time soon.

Unless you are a specialist working for a larger studio that has people specifically paid to troubleshoot and manage the toolset, the reality is that you have to deal with these kinds of problems on a regular basis.

I realize it's a stretch for a lot of sole proprietors and small shops/studios (I spent many years as one), but for your own sanity, I suggest keeping two separate systems; Stable System(s) A that is known stable that you can depend on for getting work out the door, and Development System B for trying out new tools and configurations. DO NOT MESS WITH SYSTEM A UNLESS ABSOLUTELY NECESSARY. As System A starts to become too out of date for practical use, spend some time messaging System B for maximum stability. Once you are reasonably satisfied with System B's stability (you will never be perfectly satisfied), retire system A and start building a new development system. Development System B becomes your new Productions System A. Repeat the whole process every few years.

It's really not too difficult if you build this into your regular tools update process. Essentially, you always depend on "last year's model," or more likely, two or three years' ago model for project budgeting, scheduling, etc. If Development System B is available to help, great, but don't depend on it or include it in you calculations or pricing. Yes, it means maintaining two systems, or at least keeping one separate dev system that you don't rely on. Depending on your tool set, it could also mean double licencing, or license juggling, but it's well worth it.

There are lots of different ways to follow this sort of philosophy. It could be as minimal as having two boot volumes on one machine, or it could be completely independent, fully functional systems with parallel licenses, etc. You have to figure out how much risk and headache you are willing to put up with when things go wrong, as they inevitably will.

My 2¢.