NVIDIA Spectrum-X Ethernet Networking Platform

NVIDIA Spectrum-X is an AI-optimized Ethernet networking platform that combines NVIDIA Spectrum switches with the BlueField-3 SuperNIC, ConnectX-7 NIC, and ConnectX-8 SuperNIC families to deliver high-bandwidth, lossless RoCE for the GPU-to-GPU compute (east-west) network. NVIDIA Network Operator provides the Kubernetes side: discovering the NICs, configuring rails, and exposing them to pods as schedulable resources.

Spectrum-X Multiplane is the Spectrum-X capability that splits each SuperNIC across two or more independent network planes — enabling Ethernet to scale from thousands to hundreds of thousands of GPUs in a flat, two-tier topology, with improved performance and resiliency over single-plane networks. Network Operator exposes it through the multiplaneMode field on NicConfigurationTemplate, with Hardware Multiplane (hwplb) and Software Multiplane (swplb) variants alongside the single-plane mode (none). On multiplane platforms, Hardware Multiplane is the mode Spectrum-X RA 2.3 recommends; Software Multiplane is selected only for software-based multiplane deployments.

On multiplane platforms (B300, GB300), Hardware Multiplane is the recommended mode and DOCA xPlane is a required component — it manages plane failover inside OVS-DOCA. Single-plane deployments on BlueField-3 SuperNIC do not use it. See Architecture and Components.

Spectrum-X Kubernetes deployments fall into three network architectures, distinguished by the number of planes per rail and the load-balancing mechanism:

Each Spectrum-X Reference Architecture version is supported by a specific Network Operator release:

Through Network Operator 26.4.x, spectrumXOptimized.version selected one of a fixed set of RA versions built into the NIC Configuration Operator image. From 26.7.0, NIC tuning ships as a Spectrum-X profile — a versioned YAML document published for each Reference Architecture and applied to the cluster as a labeled ConfigMap — and version names that ConfigMap. The operator carries no built-in profiles, so every Spectrum-X deployment applies a profile before configuring NICs, and tuning can be revised without a new operator release. See Spectrum-X NIC Configuration for details.

Network Operator drives Spectrum-X setup through a small set of resources. Each one is documented on the page that owns it:

For Dynamic Resource Allocation workflows, the upstream Kubernetes ResourceClaimTemplate binds Pod requests to specific GPU and VF combinations — see DRA SR-IOV Driver.

For supported platforms, NICs, switches, and OS combinations, see the NVIDIA Spectrum-X Solution Stack documentation. For the Kubernetes and OS matrix and component versions, see Platform Support.