"A graphics processing unit (GPU) is a specialized electronic circuit designed to rapidly manipulate and alter
memory to accelerate the creation of images in a frame buffer intended for output to a display device. GPUs are
used in embedded systems, mobile phones, personal computers, workstations, and game consoles. Modern GPUs are very
efficient at manipulating computer graphics and image processing, and their highly parallel structure makes them
more efficient than general-purpose CPUs for algorithms where the processing of large blocks of data is done
in parallel. In a personal computer, a GPU can be present on a video card, or it can be embedded on the
motherboard or—in certain CPUs—on the CPU die."
Quicker turnaround times for renders can be dependent on many factors and adding GPU's may not do the trick alone, so it's important for you to research your particular rendering needs. What package are you using to create the images or the scene, what kind of data are you rendering (animation, textures, etc.), and what is your desired output? Research GPU benefits against those criteria before getting started.
Qube can manage GPU rendering in the following packages:
- Cinema 4D through their native support of Radeon ProRender
More package support is in progress, so please be sure to email us if you have needs for a software package not listed.
|One benefit of using Qube for managing GPU renders are our 'non-specific GPU reservations.' Qube can track GPUs like just another resource to be reserved and scheduled. Instead of the user having to specify that they want to 'run on GPUs 1 and 2,' users can configure 'each copy of this job uses 2 GPUs.' Qube will start the job on workers that have any two GPUs free. And with Qube's random GPU assignment, Qube can alternate jobs from busy processor to open processor to moderate the heat load on the hardware therefore helping to potentially extend it's life.|
HARDWARE AND DRIVERS
If you are researching the ability to use GPU's for rendering on Windows, there is an important point to take note of. NVIDIA has two different GPU drivers for Windows, which can have an effect on how your Qube worker runs jobs. NVIDIA offers the Tesla Compute Cluster (TCC) driver and the WDDM driver. The TCC drivers will only work with the professional series cards like the Quadro and Tesla products and are not available for the GeForce GPU gaming series cards. This means you will not be able to use your machines with GeForce cards as part of your available GPU render farm.
NVIDIA drivers for Linux do not have this differentiation. The TCC driver allows for GPU rendering to be performed by a system-level service without a user logged into the host but prohibits connecting a video display to the GPU. On the other hand, the WDDM drivers, typically used when a GPU is installed in an artist's workstation, require that a user be logged into a host to run a GPU render. Your desktops must have a professional grade card and use the TCC driver to render as a system-level service.
Let's say you are not looking to render GPU's locally because you have very specific, short term needs for GPU rendering. In that case, you are most likely evaluating cloud providers. Qube works well with and has been used in production with every public cloud provider, but we have partnerships with both Google Cloud Platform (GCP) and Microsoft Azure. We work closely with both companies, are able to connect you with them, collaborate together to get companies rendering fast and efficiently in the cloud, and we are working on proof of concept integrations with each. Please do let us know if you have interest in rendering in the cloud with top class providers, Microsoft Azure Cloud or Google Cloud Platform.