Using GPUs with Legacy Engines To use GPUs with legacy engines, you must create a custom CUDA-capable engine image. Custom CUDA-capable Engine ImageSite Admins: Add the Custom CUDA Engine to your Cloudera Machine Learning DeploymentAfter you create a custom CUDA-capable engine image, you must add the new engine to Cloudera Machine Learning. Project Admins: Enable the CUDA Engine for your ProjectYou can make the CUDA-capable engine the default engine for workloads within a particular project.Testing GPU SetupUse these code samples to test that your GPU setup works with several common deep learning libraries. The specific versions of libraries depend on the particular GPU used and the GPU driver version. You can use this testing for GPU setup using ML Runtimes or Legacy Engines.