NVIDIA Mission Control Feature Support#
Platform Support Matrix#
The following table lists feature support in NVIDIA Mission Control 2.5. Vera Rubin NVL72 support is early access.
Capability |
DGX B200 * |
DGX B300 * |
DGX GB200 |
OEM GB200 NVL72 |
DGX GB300 |
OEM GB300 NVL72 |
DGX Vera Rubin NVL72 |
OEM Vera Rubin NVL72 |
|---|---|---|---|---|---|---|---|---|
Rack Provisioning and Management Software |
||||||||
Base Command Manager |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Workload Managers |
||||||||
Slurm |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Run:ai |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
No |
No |
Autonomous Resiliency Engine (ARE) |
||||||||
Autonomous Hardware Recovery |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Autonomous Job Recovery |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Telemetry and Observability |
||||||||
Observability Stack (Prometheus, Loki, Alloy, and Grafana) |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Opinionated Grafana Visualizations (JSONs) |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Mission Control Grafana Dashboards |
Yes |
Beta |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Network Management and Monitoring |
||||||||
NetQ (Formerly NMX-M) |
N/A |
N/A |
Yes |
Yes |
Yes |
Yes |
No |
No |
Minerva (Beta, Early Preview) |
N/A |
Yes |
N/A |
Not specified |
Not specified |
Not specified |
Yes |
Yes |
Mission Control LaunchPad |
||||||||
Mission Control LaunchPad |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Yes |
Management and Control Plane Hardware Requirements |
||||||||
Control Plane |
10 nodes (x86 only) |
10 nodes (x86 only) |
10 nodes; Arm required [1] |
10 nodes; Arm recommended [2] |
10 nodes; Arm required [1] |
10 nodes; Arm recommended [2] |
10 nodes; Arm required [1] |
10 nodes; Arm recommended [2] |
Note
* HGX B200 and HGX B300 systems from OEMs are not supported.
Not specified indicates that the release support matrix does not specify Minerva support for that platform.
Feature Definitions#
- Base Command Manager
Base Command Manager (BCM) 11 provides cluster deployment, management, and monitoring.
- Slurm
Workload scheduling and resource management for batch jobs.
- Run:ai
Workload orchestration and GPU resource management for AI workloads.
- Autonomous Hardware Recovery
Hardware health monitoring, diagnostics, and recovery workflows.
- Autonomous Job Recovery
Job monitoring and recovery workflows for failed jobs.
- Observability Stack
Prometheus, Loki, Alloy, and Grafana provide metrics, logs, and visualization.
- Opinionated Grafana Visualizations (JSONs)
Grafana visualization definitions provided as JSON files.
- Mission Control Grafana Dashboards
Grafana dashboards for NVIDIA Mission Control monitoring and telemetry visualization.
- NetQ (Formerly NMX-M)
NVLink fabric monitoring and management for GB200 and GB300 NVL72 systems.
- Minerva
An AI operations layer that correlates network fabric telemetry with GPU, interconnect, scheduler, and workload signals. It supplements the Observability Stack.
- Dynamic Power Software
Power management software available separately as a pre-release beta through a scoped engagement. Refer to the availability details that follow.
- Mission Control LaunchPad
An entry point for Mission Control services and workflows.
- Control Plane
Management infrastructure, including node counts and processor architecture requirements.
Architecture Requirements#
The control plane requires 10 nodes for each platform listed in the support matrix. Architecture requirements vary by platform.
DGX B200 and DGX B300#
Use x86 nodes for the control plane. HGX B200 and HGX B300 systems from OEMs are not supported.
DGX GB200, DGX GB300, and DGX Vera Rubin NVL72#
Arm nodes are required for BCM head nodes and Slurm nodes. Exceptions are permitted for x86 BCM head nodes.
OEM GB200, OEM GB300, and OEM Vera Rubin NVL72#
Arm nodes are recommended for Slurm nodes. The rest of the control plane can use x86 nodes.
Feature Availability#
The following capabilities are supported across all platforms in the matrix:
Base Command Manager
Slurm
Autonomous Hardware Recovery
Autonomous Job Recovery
Observability Stack
Opinionated Grafana Visualizations (JSONs)
Mission Control LaunchPad
The following capabilities have platform-specific availability:
Run:ai is supported on B200, B300, GB200, and GB300 platforms. It is not supported on Vera Rubin NVL72 platforms in NVIDIA Mission Control 2.5.
Mission Control Grafana Dashboards are available across all platforms. DGX B300 dashboards are in beta.
NetQ is supported on DGX and OEM GB200 and GB300 NVL72 systems. It is not applicable to DGX B200 or DGX B300 and is not supported on Vera Rubin NVL72.
Minerva is available in beta for DGX B300 and DGX and OEM Vera Rubin NVL72. It is not applicable to DGX B200 or DGX GB200. Support for OEM GB200, DGX GB300, and OEM GB300 is not specified.
Minerva and Dynamic Power Software (DPS) are separate pre-release beta components. They are not included in the NVIDIA Mission Control 2.5 release and are available only through a scoped engagement. Contact your NVIDIA account manager to participate.
DPS 0.9.0 supports GB200 NVL72 and GB300 NVL72, with early access support for Vera Rubin NVL72. Known issues prevent its use on DGX B200 and DGX B300.
Platform Comparison#
The following table summarizes control plane differences across platform groups.
Platforms |
Control Plane Nodes |
Architecture Requirements |
|---|---|---|
DGX B200 and DGX B300 |
10 |
x86 only |
DGX GB200, DGX GB300, and DGX Vera Rubin NVL72 |
10 |
Arm required for BCM head nodes and Slurm nodes; exceptions permitted for x86 BCM head nodes |
OEM GB200, OEM GB300, and OEM Vera Rubin NVL72 |
10 |
Arm recommended for Slurm nodes; the rest can use x86 nodes |
Additional Resources#
For hardware specifications and control plane requirements, refer to the NVIDIA Mission Control Hardware Requirements Guide.