Overview

The BMC node enables remote power cycling, board environment monitoring, NVIDIA® BlueField® SoC temperature monitoring, board power and consumption monitoring, and individual interface resets. The BMC also supports the ability to push a bootstream to the BlueField. It is recommended to manage NVIDIA® BlueField® using Redfish commands. However, IPMI commands and sysfs monitoring infrastructure are available as well .

The procedures described in this manual assume that you have already installed and powered on your device according to the instructions in BlueField's specific hardware guide.

• Support for IPMI 2.0 (v1.1) Standards

  • Thermal control – access to all relevant temperature sensors, fan control

  • System management – power state control, power on/off, reboot/reset

  • Environmental monitoring – voltage/current/power

  • Serial over LAN (SOL)

  • RMCP/RMCP+

  • Event log management

  • Event alerting

  • VLAN support

• Support for DMTF Standards

  • Redfish specification (DSP0266)

  • Network Controller Sideband Interface (NC-SI) Specification (DMTF DSP0222)

• Support for BMC image update

The NVIDIA® BlueField® DPU is a data center infrastructure on a chip that combines a high-speed networking interface with powerful, software-programmable Arm cores, enabling breakthrough networking, storage, and security performance. The BlueField DPU offloads, accelerates, and isolates a broad range of software-defined infrastructure services which traditionally ran on the host's CPU, overcoming performance and scalability bottlenecks, and eliminating security threats in modern data centers.

BlueField DPUs transform traditional computing environments to secure and accelerated data centers, allowing organizations to efficiently run data-driven, cloud-native applications alongside legacy applications. By decoupling the data center infrastructure from business applications, BlueField DPUs enhance data center security, streamline operations, and reduce total cost of ownership.

The BlueField DPU contains a programmable CPU based on Arm cores, a state-of-the-art NVIDIA® ConnectX®, and an enhanced set of security, storage, and networking accelerators that can be configured to perform multiple software-defined, hardware-accelerated functions. With a BlueField DPU, a software-defined network, and/or software-defined storage solution can be deployed and offloaded from the main host CPU in the server. Similarly, other dedicated services (e.g., distributed firewall, deep packet inspection, malware detection) can run on the BlueField DPU and can be accelerated with zero CPU overheads.

The BlueField DPU resembles a server embedded within the server itself, creating a secure environment where an infrastructure stack can operate independently from the primary (i.e., host) CPU, effectively isolating it from the untrusted tenant applications.

This is the recommended mode for utilizing the DPU in which software running on the host CPU has no direct access to the DPU. For instance, in a scenario where a cloud service provider is responsible for managing both networking and storage in a cloud infrastructure stack, it can establish an isolated environment within the DPU.

The NVIDIA® BlueField® SuperNIC is the world’s most advanced network accelerator, designed for supercharging hyperscale generative AI workloads. It delivers deterministic, isolated performance, with secure cloud multi-tenancy. Featured on the Spectrum-X networking platform, NVIDIA integrates BlueField-3 SuperNICs across its accelerated systems to enable peak AI workload efficiency. Powered by the NVIDIA DOCA software, the SuperNIC offers up to 400Gb/s connectivity between GPU servers, with features like RoCE adaptive routing, direct data placement (DDP), and programmable congestion control. With its unique HHHL form factor and low-power platform, the BlueField-3 SuperNIC fits most enterprise-class servers.