Abstract

This support matrix provides a single view into the supported software and specific versions that are supported today on NVIDIA Base Command.

Base Command Support Matrix

Note: The deep learning framework container packages follow a naming convention that is based on the year and month of the image release. For example, the 21.02 release of an image was released in February 2021.
NVIDIA Optimized Frameworks
PyTorch >= 20.09 Getting Started Contents
  • Python 3

  • APEX

  • MLNX_OFED

  • OpenMPI

  • Nsight Compute

  • Nsight Systems

  • TensorBoard

  • TensorRT

  • DALI

  • MAGMA

  • DLProf

  • PyProf

  • Tensor Core optimized examples:

    • ResNeXt
    • SE-ResNext
    • TransformerXL
    • Jasper
    • BERT
    • Mask R-CNN
    • Tacotron 2 and WaveGlow
    • SSD300
    • Neural Collaborative Filtering (NCF)
    • ResNet50
    • GNMT
  • Jupyter and JupyterLab:

    • Jupyter Client
    • Jupyter Core
    • Jupyter Notebook
    • JupyterLab
    • JupyterLab Server
    • Jupyter-TensorBoard
    • PyProf
Tensorflow >=20.09 Getting Started Contents
  • Python 3
  • Horovod
  • OpenMPI
  • TensorBoard
  • MLNX_OFED
  • TensorRT
  • DALI
  • Nsight Compute
  • Nsight Systems
  • Jupyter and JupyterLab:
    • Jupyter Client0
    • Jupyter Core
    • Jupyter Notebook
    • JupyterLab
    • JupyterLab Server
    • Jupyter-TensorBoard
KALDI >= 20.09 Getting Started Contents
  • Python 3
  • MLNX_OFED
  • OpenMPI
  • Nsight Compute
  • Nsight Systems
  • TensorRT
Base Container Image
Container OS Ubuntu 18.04/Ubuntu 20.04/UBI8
CUDA Base Image ( cuBLAS, cuDNN, NCCL )

NVIDIA CUDA 11.4.2 (CUDA 11.x supported through compatibility on R450)

NVIDIA Video Codec SDK Video Codec SDK 11.1 video-codec-sdk
Vulkan SDK ( Vulkan and OpenGL ) SDK 1.2.133 Vulkan Container
Optix (not containerized) 7.1.0 About the NVIDIA Optix Ray Tracing Engine

DGX

DGX System DGX A100 640 GB About the DGX A100
Operating System DGX OS 5 Release Notes
NVIDIA Driver 450.142.00 Release Notes
GPU NVIDIA A100 About the NVIDIA A100

Getting Support from the Base Command Support Forum

For more information regarding model or application support, visit the NVIDIA Base Command Support Forum.