ML Runtimes NVIDIA GPU Edition
The NVIDIA GPU Edition Runtimes are built on top of NVIDIA CUDA docker images. CUDA is a parallel computing platform and programming model developed by NVIDIA for general computing on graphical processing units (GPUs). With CUDA, developers can dramatically speed up computing applications by harnessing the power of GPUs.
Cloudera Machine Learning supports GPUs using particular ML Runtime Editions. NVIDIA GPUs together with related specialized software can be utilized using the NVIDIA GPU Edition.
Version information:
| Version | Additional software | Tested OS | Tested Driver Version |
|---|---|---|---|
| 2021.09 | CUDA 11.4 | RHEL 7.6 | NVIDIA driver 460.56 |
Compatibility information:
| Runtime Version | CUDA Version | Kernels | Editors | Base Image |
| 2021.02 | CUDA 11.1 | Python 3.7 and Python 3.8 | Workbench and JupyterLab |
ML Runtimes inherit the compatibility requirements for NVIDIA CUDA. For compatibility information, visit the following pages:
