Using GPU Resources on CircleCI


Our GPU-enabled cloud machines are currently only available to users on a Scale or Custom Annual plan as outlined on the following pages:

Users who try to use a GPU resource class on an ineligible CircleCI plan will be met with a message similar to the one below:


Note: GPU Builders are not currently supported in Server 3.x and 4.x


To use our cloud GPU resources, the organization using these in their configuration needs to be on the Scale or Custom Annual Plan.


Selecting GPU Resource Classes

We offer resource classes on both Windows and Linux with GPUs available which all include an Nvidia Tesla GPU. The available GPU resource classes can be found in the following section of our documentation:

Using a Self-Hosted Runner

For organizations not on the required plans, we suggest trying out a machine runner with GPU resources available on that machine/instance. In this case, the user would provide a GPU-enabled machine locally or perhaps an instance on a third-party service (Such as AWS, GCP, or any other cloud provider that offers these). This means that while the user provides the hardware, CircleCI will run all of your jobs, and tests similar to using a resource class provided by CircleCI.

More details about our runner offering can be found in the section of our documentation linked below:


Additional Resources

Was this article helpful?
0 out of 0 found this helpful



Article is closed for comments.