List of NVIDIA-Certified Systems

Reference Configurations

Reference Configurations are NVIDIA-Certified server configurations that form the foundation of NVIDIA Reference Architectures such as NCP Reference Configurations and Enterprise Reference Configurations.

Reference Configurations include PCIe Optimized and HGX platforms that follow prescriptive design patterns to ensure optimal performance when deployed in a cluster environment.

Systems previously tested in Reference Configurations with NVIDIA L40S and L40 GPUs are certified for NVIDIA OVX™. Systems previously tested in Reference Configurations with NVIDIA B3140H or NVIDIA ConnectX®-8 SuperNICs for East-West data connectivity are Spectrum-X™ Compatible.

Solutions based on NVIDIA Enterprise Reference Architectures are available from NVIDIA’s global partners. Contact your partner representative for details.

Learn more about Enterprise Reference Architectures in the NVIDIA Enterprise Reference Architectures documentation.

PCIe Optimized Reference Configuration

PCIe Optimized Reference Configurations are NVIDIA-Certified scale-out compute nodes that use PCIe GPUs and high-bandwidth NVIDIA networking for compute and graphics workloads. Each configuration name follows the pattern CPU-GPU-Network Adaptor-Average GPU Network Bandwidth. For example, 2-4-3-200 represents two CPUs, four PCIe GPUs, three network adaptors, and an average network bandwidth of 200 Gb/s.

The PCIe Optimized 2-2-3-400 configuration supports two PCIe GPUs and three network adaptors. The PCIe Optimized 2-4-3-200 configuration supports four PCIe GPUs and three network adaptors. The PCIe Optimized 2-8-5-200 configuration supports eight PCIe GPUs and five network adaptors.

The supported PCIe GPUs and network adaptors are listed in the NVIDIA Enterprise Reference Architecture Whitepaper.

PCIe Optimized Reference Configuration 2-2-3-400

PCIe Optimized Reference Configuration 2-4-3-200

PCIe Optimized Reference Configuration 2-8-5-200

HGX Reference Configuration

HGX Reference Configurations are NVIDIA-Certified scale-up server platforms for large-scale AI training, inference, and other accelerated computing workloads. Configuration names follow the pattern CPU-GPU-Network Adaptor-Average GPU Network Bandwidth. For example, 2-8-9-400 represents two CPUs, eight GPUs, nine network adaptors, and an average network bandwidth of 400 Gb/s.

The 2-8-9-400 configuration supports NVIDIA HGX H100/H200 systems. The 2-8-9-800 and 2-8-10-800 configurations support NVIDIA HGX B200/B300 systems with an average network bandwidth of 800 Gb/s. Spectrum-X Compatible Reference Configurations are tested with NVIDIA networking devices such as B3140H and ConnectX-8. Supported HGX GPUs and network adaptors are listed in the NVIDIA Enterprise Reference Architecture Whitepaper.

HGX Reference Configuration: 2-8-9-400

HGX Reference Configuration: 2-8-9-800

HGX Reference Configuration: 2-8-10-800

Reference Configuration (Inference Only)

These systems support inference-only deployments with RTX PRO 6000 Blackwell Server Edition and RTX PRO 4500 Blackwell Server Edition GPUs in 2-8-1 and 2-4-1 configurations.

These configurations do not define an East-West network and use only DPU-based North-South connectivity.

Reference Configuration (Inference Only) 2-8-1

Reference Configuration (Inference Only) 2-4-1

Systems Certified for NVIDIA Confidential Computing

NVIDIA-Certified systems testing for Confidential Computing validates robust application performance with Confidential Computing. These systems provide hardware-based protection for AI workloads and sensitive data assets while maintaining optimal NVIDIA acceleration. Learn more about NVIDIA Confidential Computing.

NVIDIA Confidential Computing

• Dell Technologies PowerEdge R770 Rack Server with RTX PRO 6000 Blackwell Server Edition

• Dell Technologies PowerEdge R7715 Rack Server with RTX PRO 6000 Blackwell Server Edition

• Dell Technologies PowerEdge XE7740 Rack Server with RTX PRO 6000 Blackwell Server Edition

• Dell Technologies PowerEdge XE7745 Rack Server with RTX PRO 6000 Blackwell Server Edition

• HPE ProLiant Compute DL380a Gen12 with RTX PRO 6000 Blackwell Server Edition

• Supermicro AS-4126GS-NB3RT-LCC with HGX B300

• Supermicro AS-4126GS-NBR-LCC with HGX B200

• Supermicro AS-A126GS-TNBR with HGX B200

• Supermicro SYS-422GA-NBRT-LCC with HGX B200

• Supermicro SYS-422GS-NB3RT-ALC with HGX B300

• Supermicro SYS-422GS-NB3RT-LCC with HGX B300

• Supermicro SYS-822GS-NB3RT with HGX B300

• Supermicro SYS-822GS-NBRT with HGX B200

Data Center Servers

The following systems have been validated by NVIDIA as NVIDIA-Certified Systems - Data Center Servers.

The NVIDIA GPUs listed in the following table are certified for Data Center Servers across all generally available GPU versions. For example, any system validated for the A100 GPU is certified for both the A100 40 GB and A100 80 GB versions.

NVIDIA AI Enterprise is supported on NVIDIA-Certified Systems across bare metal, virtualized, and cloud deployments. For a complete list of supported software platforms, refer to the NVIDIA AI Enterprise Infrastructure Support Matrix.

Each NVIDIA-Certified System is tested with supported NVIDIA GPUs to validate the performance and reliability of the combined system.

Data Center Servers

Note:

¹ GPU power is capped based on system capabilities. Contact the partner for details.

² Modular system with built-in OEM networking.

DGX Spark / Grace Blackwell GB10 Systems

The following systems have been validated by NVIDIA as NVIDIA-Certified Systems - NVIDIA DGX™ Spark / Grace Blackwell GB10 systems. Delivering the power of Grace Blackwell in a desktop-friendly size, NVIDIA-Certified partner systems built with NVIDIA GB10 GPUs are ideal for AI developers, researchers, and data scientists.

DGX Spark / Grace Blackwell GB10 Systems

DGX Station / Grace Blackwell GB300 Systems

The following systems have been validated by NVIDIA as NVIDIA-Certified Systems - NVIDIA DGX Station™ / Grace Blackwell GB300 systems. These deskside AI supercomputers let you develop, fine-tune, and run inference with large AI models and long-running AI agents locally. You can then deploy them to the data center or cloud.

DGX Station / Grace Blackwell GB300 Systems

NVIDIA Arm CPU Systems

The following systems have been validated by NVIDIA as NVIDIA-Certified Systems - NVIDIA Arm CPU Systems and are listed in the following table.

NVIDIA Arm systems use the NVIDIA Grace™ CPU architecture, which is designed for hyperscale, enterprise, and high-performance computing (HPC) applications. The Grace CPU uses NVIDIA NVLink™-C2C to enable access by the GPU at seven times the bandwidth of a traditional PCIe connection. The NVIDIA Grace Hopper™ Superchip pairs a power-efficient, high-bandwidth NVIDIA Grace CPU with an NVIDIA H100 Tensor Core GPU to maximize the capabilities for accelerated computing and generative AI workloads.

NVIDIA Arm CPU Systems

Note:

^* Tested as a standalone system.

Workstations

The following systems have been validated by NVIDIA as NVIDIA-Certified Systems - Workstations.

NVIDIA-Certified workstations provide the hardware foundation of Data Science Workstations that combine optimized hardware with a data science software stack built on NVIDIA CUDA-X™ AI. They use enterprise-class GPUs to accelerate data science workloads and are validated for optimal performance, reliability, and compatibility with NVIDIA software and services.

NVIDIA-Certified systems are tested with each supported NVIDIA GPU to ensure that they can work together to provide the highest level of performance and reliability.

Workstations

Mobile Workstations

The following systems have been validated by NVIDIA as NVIDIA-Certified Systems - Mobile Workstations.

NVIDIA-Certified workstations provide the hardware foundation of Data Science Workstations that combine optimized hardware with a data science software stack built on NVIDIA CUDA-X AI. They use enterprise-class GPUs to accelerate data science workloads and are validated for optimal performance, reliability, and compatibility with NVIDIA software and services.

NVIDIA-Certified systems are tested with each supported NVIDIA GPU to ensure that they can work together to provide the highest level of performance and reliability.

Mobile Workstations

High-Density VDI Servers

The following systems have been validated by NVIDIA as NVIDIA-Certified Systems - High-Density VDI Servers.

NVIDIA-Certified systems are tested with each supported NVIDIA GPU to ensure that they can work together to provide the highest level of performance and reliability.

High-Density VDI Servers

Edge Systems

The following servers and workstations have been validated by NVIDIA as NVIDIA-Certified Systems - Edge Systems.

NVIDIA AI Enterprise is supported when used on NVIDIA-Certified systems in bare metal deployments, virtualized deployments, and in the cloud. The supported bare metal, virtualized, and cloud software platforms are listed in the NVIDIA AI Enterprise Infrastructure Support Matrix.

NVIDIA-Certified systems are tested with each supported NVIDIA GPU to ensure that they can work together to provide the highest level of performance and reliability.

Enterprise Edge servers are tested for remote systems management using the industry-standard Redfish API.

Industrial Edge workstations are tested at elevated ambient temperatures to demonstrate their ability to function effectively outside a typical data center environment. Contact your hardware provider for the specific NVIDIA GPUs supported for this use case.

Edge Systems

IGX Systems

The following systems have been validated by NVIDIA as NVIDIA-Certified Systems - IGX Systems.

NVIDIA IGX™ is an industrial-grade, edge AI platform that combines enterprise-level hardware, software, and support. It is purpose-built for industrial and medical environments, delivering powerful AI compute, high-bandwidth sensor processing, enterprise security, and functional safety. The platform also comes with NVIDIA AI Enterprise and up to 10 years of support, so you can confidently deliver AI safely and securely to support human and machine collaboration.

NVIDIA-Certified IGX systems are equipped with an integrated NVIDIA Arm GPU and can be optionally configured with a discrete NVIDIA GPU.

NVIDIA IGX Systems