Overview

This guide provides information on how to troubleshoot and resolve V-Ray GPU crashes causing the following error messages:

CUDA error 719, Cuda error 700, Could not release device buffer

Optix error, Error Code 7900

Root causes and solutions

CUDA and Optix errors may be encountered during production or interactive rendering with the V-Ray GPU render engine. They may be caused by several different factors, more information about each of them could be found below.

Note: Please follow the troubleshooting steps below in order starting from Step 1.

1. Using a not verified GPU driver version.

Using the recommended driver is essential for a stable experience with V-Ray GPU.
You can check the currently installed driver's version by going to:
NVidia Control Panel > Help > System information.
(Here is a video showing you how to check it)

The current recommended driver can be found at the top of the V-Ray GPU website here.

For the user's convenience, V-Ray GPU performs a graphics driver check at the beginning of the render and prints a warning message in the V-Ray Messages Log window if the current driver is different from the recommended one.

Often newer versions of drivers work just fine but for the best V-Ray GPU experience, we advise using the recommended driver.

Solution: Install recommended GPU driver.

2. Hardware malfunction

The graphics card installed on the machine might work fine with most everyday tasks, but it could be unstable when heavily utilized with software like V-Ray GPU. To identify if this is the case, we recommend performing GPU (CUDA) stress tests with V-Ray Benchmark. In addition, you can also perform tests with software like FurMark, OctaneBench, RedshiftBench and etc.

If there is an issue with V-Ray Benchmark and other benchmarks, this means that the CUDA error is related to hardware or OS malfunction. V-Ray is not the root cause here, it only triggers the error.

Solution: Contact your hardware supplier or OS support to assist you on this matter.

Additional information regarding hardware malfunctioning can be found here.

3. Scene-related issue

The next step you need to check is whether the error occurs with specific scenes/projects only. To narrow down the possible cause try to render other projects/scenes and see if the same error occurs again or not. Try to also render a new empty file, this is the easiest way to find out whether the issue is general and occurs with every scene, or it's scene-dependent.

- If the issue reproduces with every single scene file (including a new empty file), then the reason for the error is either the driver (1) or the hardware (2).

- If the issue reproduces with a specific scene only, then the issue could be a bug or an insufficient memory issue. Please proceed with step 4 to ensure the issue is not caused by insufficient GPU memory, and if that's not the case follow the solution below.

Solution: Submit a new support request from here by including the following information:

Scene file (including assets) that reproduces the error.
VRayLog.txt (can be found in %temp% folder).
Steps to reproduce the issue (if error occurs with specific steps only).

4. GPU Memory / VRam

Another common reason for Cuda errors is insufficient GPU memory. Please note that sometimes you may encounter crashes even though the GPU memory is not fully utilized (100%). This is happening when V-Ray requests more memory from the GPU driver than the currently available free one. For example, if the current free GPU memory is 4GB and V-Ray requests 6GB it will crash even before the memory is fully utilized.

To find out whether the issue is related to the GPU memory, start removing objects from the scene and see if the error will go away or try to render the scene on another machine with a GPU device that has more memory.

Solution 1: If more than one GPU device is available on the machine, render the scene only on the device with the most GPU memory available.

Solution 2: Optimize the scene file to use less GPU memory. You can find out how to do it in this article.

Solution 3: Upgrade your GPU device with another one with more GPU memory.