NGC | Catalog
CatalogContainersNVIDIA MLPerf Inference

NVIDIA MLPerf Inference

For copy image paths and more information, please view on a desktop device.
Logo for NVIDIA MLPerf Inference


MLPerf Inference containers are base containers for people interested in NVIDIA's MLPerf Inference submission results



Latest Tag



September 6, 2023

Compressed Size

16.51 GB

Multinode Support


Multi-Arch Support


mlpinf-v3.1-cuda12.2-cudnn8.9-aarch64-ubuntu22.04-public (Latest) Scan Results

Linux / arm64

MLPerf Inference NVIDIA-Optimized Implementations

MLPerf Inference is a benchmark suite for measuring how fast systems can run models in a variety of deployment scenarios. MLPerf Inference provides the base containers to enable people interested in NVIDIA’s MLPerf Inference submission to reproduce NVIDIA’s leading results. Containers included are sorely for benchmarking purposes and should not be used in any production environment.

Getting Started

For details of how to reproduce NVIDIA's results, please visit ML Commons github page and check NVIDIA's submission repo's readme file.


The user license has been include under the container's root directory as /NVIDIA_MLPerf_Evaluation_License. By downloading this container, you agree to follow all the requirements stated in the EULA.


MLCommons offical webpage: