NVIDIA BLUEFIELD
​​​​​​​DATA PROCESSING UNITS

NVIDIA BlueField-3 is the first DPU with 400 Gb/s/NDR and high-speed processing for software-defined networking, storage, and cybersecurity. BlueField-3 combines high-performance computing, high-speed networking, and extensive programmability to deliver software-defined, hardware-accelerated solutions for the most demanding workloads. From accelerated AI computing to hybrid cloud to cloud-native supercomputing and 5G wireless networking, BlueField-3 pushes the boundaries of what is possible.

NVIDIA BlueField-2 is a new breed of DPU, delivering innovative acceleration, security, and efficiency for any host. The BlueField-2 on-a-chip data center infrastructure combines the performance of the ConnectX-6 Dx with programmable arm cores and hardware offloads for software-defined storage, networking, security, and management workloads.

The NVIDIA BlueField-2 DPU product line also includes a controller optimized for NVMe storage that delivers superior performance and reduced TCO when managing backend NVMe storage, all-flash arrays (AFA), and hyper-converged systems.

The world's first AI-powered DPU: NVIDIA BlueField-2X This advanced DPU has all the key features of BlueField-2, but enhanced with the AI functionality of the NVIDIA Ampere GPU. NVIDIA's third-generation Tensor compute units enable AI to be used for real-time security and management analytics, such as abnormal traffic detection that could indicate theft of confidential data, host introspection to detect malicious activity, and dynamic security orchestration with automated response. BlueField-2X enables high-speed analysis of encrypted traffic, connection tracking and adaptive telemetry.

NVIDIA EGX™ converged accelerators combine the powerful performance of NVIDIA Ampere architecture with the enhanced security and latency reduction features of NVIDIA's SmartNIC and DPU technologies. Hospitals, retail, warehouses, factories, airports, and other organizations can use EGX converged accelerators to create faster, more efficient, and more secure AI systems in enterprise data centers and in the periphery.

The NVIDIA DOCA SDK enables developers to build tomorrow's data center infrastructure by developing software-defined, cloud-native, DPU-accelerated services to increase performance and security.

NVIDIA Certified Systems™ provide true multi-node scaling with NVIDIA Ampere GPUs and BlueField DPUs, optimized and tested to run advanced AI workloads with built-in security solutions such as in-line encryption, secure boot, and air-gap isolation. The NVIDIA DOCA SDK, available in the NVIDIA NGC™ catalog, provides a convenient container environment for third-party vendors to leverage advanced data center acceleration on NVIDIA-certified systems to develop, certify, and distribute applications to end customers.

NVIDIA Morpheus is an AI application framework that provides developers with a new level of data center security through a highly optimized AI pipeline and pre-trained AI capabilities.

The Morpheus framework provides highly scalable and adaptive data collection, pre-trained AI models, and policy control capabilities that enable developers to detect and remediate security threats.