NVIDIA DGX Cloud Benchmarking

How NVIDIA DGX Cloud Benchmarking Works

NVIDIA DGX Cloud Benchmarking analyzes real-time metrics across GPU configurations and environments, with ready-to-use benchmarking templates and on-demand benchmarking for custom workloads. Baseline performance results for comparison are provided via interactive dashboards.

Specifications:

 - Scaling analysis from 8 to 2,048 GPUs
 - Precision comparisons: FP8 versus BF16
 - Support for popular AI frameworks and models
 - Performance data across NVIDIA NeMo? framework versions

Example of improved metrics gained over time derived from NVIDIA DGX Cloud Benchmarking results.

Get Started With NVIDIA DGX Cloud Benchmarking

Benchmark AI Workloads

Understand end-to-end platform performance and learn best practices for cluster sizing and workload optimization with NVIDIA DGX Cloud Benchmarking.

Try Now

Benchmarking Recipes

Deploy ready-to-use benchmarking templates across any cloud platform. Leverage NVIDIA’s performance baselines to compare expected performance across precisions and scales.

View Recipes

NVIDIA DGX Cloud Benchmarking Starter Kits

See how well your own environment performs: log into NGC; select a model; download the container, benchmarking recipes, and dataset scripts; and launch to obtain throughput results.

NVIDIA DGX Cloud Benchmarking Learning Library

Ethical AI

NVIDIA believes Trustworthy AI is a shared responsibility, and we have established policies and practices to enable development for a wide array of AI applications. When downloaded or used in accordance with our terms of service, developers should work with their supporting model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse.

For more detailed information on ethical considerations for this model, please see the Model Card++ Explainability, Bias, Safety & Security, and Privacy Subcards. Please report security vulnerabilities or NVIDIA AI Concerns here.