Introduction
The GPU device selector is used to choose the GPU engine type between CUDA and RTX, as well as devices used for GPU rendering. The GPU device selector has been reworked in V-Ray 6 with new UI and new features.
Note: changing the settings requires a restart of the V-Ray Standalone.
Standalone:
V-Ray GPU in 3Ds Max:
- Engine: chooses the Rendering device or Denoising device. The render engine can be CUDA or RTX.
- GPUs: lists all available devices for GPU rendering. If the Engine is set to CUDA, you will be able to use your CPU(s) in rendering, marked by C++/CPU. In the screenshot, machine has a Ryzen 9 5950X 16-Core.
- Rendering: selects the GPU and/or CPU devices for rendering.
- Denoising: selects the device for Denoising. This affects the Nvidia AI denoiser as well as V-Ray's Denoiser.
Note: CPU devices cannot be selected for denoising in the UI. The V-Ray Denoiser will fall back to using CPU if there are no available GPU devices selected.
Note: when multiple devices are selected for denoising, out of all the devices that match the query, only the device with the highest capability is used for the denoising process. You can set denoising to your display device marked by the asterisk (*) if you don't use this GPU for rendering (as shown on the above screenshot).
- Low priority: Sets a low GPU thread priority per device. When enabled, V-Ray tries to lessen the load on GPUs working on displaying graphics to the monitor(s) to give them more resources to complete other processes and tasks for the OS. This is done by internally using a lower value for Rays bundle size for those GPUs with attached displays (Rays per pixel is still the same for all GPUs). This can affect the overall performance, and the render speed might be reduced. It is recommended to utilize a separate GPU for the display, if possible.
Note: high-end GPUs don't need this option. It should only be used with mid to low-end GPUs if you get sluggish UI during interactive or final rendering.
New Functionality.
- Choosing a device for Denoising is possible.
- Choosing Low Priority per device is possible.
- The asterisk (*) sign next to a GPU device's name in the list means that a monitor is connected to that GPU device. If two or more monitors are connected to the GPU device, there are 2 or more asterisks next to the GPU device name. In the example screenshots below, 3 monitors connected to GeForce RTX 3090 - therefore, marked by (***).
- There is a tooltip that appears upon hovering over the asterisks and shows information on the GPU and the number of monitors connected to it:
Tips and tricks for using the GPU Device Selector.
- Being able to choose a device for denoising, improves the interactive performance as well as GPU memory usage as Denoising happens on a separate device. This could be done in the DCC's UI or in the standalone Device Select tool.
In the screenshot above, denoising is set to the RTX 3090 used for monitors, also marked by *** next to its name. The RTX 4090s are dedicated to GPU rendering. For more information about interactive rendering, please check the article Using interactive rendering in V-Ray GPU. - If you get sluggish UI during interactive or final rendering, you can use the Low Priority option.
- The asterisk (*) helps you identify the device connected to your monitors. This is useful if you have multiple identical GPUs.
- When using both CUDA and CUDAx86 modes (aka hybrid GPU+CPU or XPU) the contribution of the CPU is similar to that of adding another (typically smaller) GPU to a multi-GPU configuration. CPUs with very high core counts (e.g., an AMD Threadripper) can often rival the speed of a GPU as shown in the graph below: