NVIDIA Mission Control 2.1.0 Feature Support#

Platform Support Matrix#

The following table outlines feature support across different NVIDIA Mission Control platforms:

Capability

DGX B200/B300 *

DGX GB200

OEM GB200 NVL72

DGX GB300

OEM GB300 NVL72

NMC/BCM11

Yes

Yes

Yes

Yes

Yes

Run.ai

Yes

Yes

Yes

Yes

Yes

Autonomous Hardware Recovery

Planned

Yes

Yes

Yes

Yes

Autonomous Job Recovery

Planned

Yes

Yes

Planned

Planned

Observability

Base View-Based

Yes (Full Stack)

Yes (Full Stack)

Planned

Planned

NMX-M/NetQ

NetQ soon

Yes

Yes

Yes

Yes

Control Plane
7 – for DGX B200
10 – DGX B300
x86 Only
10 nodes
(ARM required for BCM Head Nodes and Slurm nodes, exceptions permitted for x86 head nodes)
10 nodes
(ARM recommended for Slurm nodes, rest can be x86)
10 nodes
(ARM required for BCM Head Nodes and Slurm nodes, exceptions permitted for x86 head nodes)
10 nodes
(ARM recommended for Slurm nodes, rest can be x86)

Note

* = HGX/OEM systems not supported

Feature Definitions#

NMC/BCM11

NVIDIA Mission Control with Base Command Manager version 11 support for cluster deployment, management, and monitoring.

Run.ai

Workload orchestration and GPU resource management platform integration for AI workload scheduling and resource allocation.

Autonomous Hardware Recovery

Automatic detection and recovery from hardware failures without manual intervention. Continuously monitors hardware health and performs self-healing operations.

Autonomous Job Recovery

Automatic restart and recovery of failed jobs with checkpoint support. Monitors job state and restarts workloads on recoverable failures.

Observability

Monitoring and telemetry capabilities:

  • Base View-Based: Limited monitoring capabilities through Base Command View interface

  • Full Stack: Complete system monitoring across all layers including metrics, logs, dashboards, and alerts

NMX-M/NetQ

Network monitoring and management tools:

  • NMX-M: NVLink Management Software for NVLink fabric monitoring (GB-series only)

  • NetQ: Network fabric monitoring and troubleshooting (available in v2.1.0)

Control Plane

Management infrastructure requirements including node count and architecture specifications (x86 vs ARM).

Architecture Requirements#

DGX B200/B300 Series#

DGX B200:

  • 7 control plane nodes total:

    • 2 BCM Head Nodes (x86)

    • 3 Run:AI Management Nodes (x86)

    • 2 Slurm Nodes (x86)

  • x86 architecture only

  • Traditional DGX systems in standard rack configuration

  • HGX/OEM systems not supported

DGX B300:

  • 10 control plane nodes total:

    • 2 BCM Head Nodes (x86)

    • 3 Run:AI Management Nodes (x86)

    • 3 Admin Kubernetes Nodes (x86)

    • 2 Slurm Nodes (x86)

  • x86 architecture only

  • Traditional DGX systems in standard rack configuration

  • HGX/OEM systems not supported

DGX GB200 Series#

  • 10 control plane nodes total:

    • 2 Head Nodes (x86) - Admin Control Plane

    • 3 Admin Service Nodes (x86)

    • 2 Slurm Nodes (ARM) - User Control Plane

    • 3 User Service Nodes (x86)

  • ARM architecture required for BCM Head Nodes and Slurm nodes

  • Exceptions permitted for x86 head nodes in specific configurations

  • NVL72 rack-based system with compute trays

  • Separated admin and user control planes

OEM GB200 NVL72 Series#

  • 10 control plane nodes total

  • ARM architecture recommended (but not required) for Slurm nodes

  • Rest of control plane can be x86

  • NVL72 rack-based system with compute trays

  • Autonomous Job Recovery not supported

DGX GB300 Series#

  • 10 control plane nodes total:

    • 2 BCM Head Nodes (ARM)

    • 3 Admin Kubernetes Nodes (x86)

    • 2 Slurm Nodes (ARM)

    • 3 User Kubernetes Nodes (ARM)

  • ARM architecture required for BCM Head Nodes and Slurm nodes

  • x86 exceptions permitted for head nodes in specific configurations

  • Hybrid ARM/x86 architecture

  • NVL72 rack-based system with compute trays

  • Autonomous Job Recovery in planned status

OEM GB200 NVL72 Series (Alternative)#

  • 10 control plane nodes total

  • ARM architecture recommended (but not required) for Slurm nodes

  • Rest of control plane can be x86

  • NVL72 rack-based system with compute trays

  • Autonomous Job Recovery not supported

Feature Availability by Version#

Mission Control 2.1.0 Features#

Available in v2.1.0:

  • NetQ network monitoring (DGX B300, DGX GB300, all GB200 variants)

  • Full Observability Stack (DGX GB200, all GB variants)

  • Autonomous Hardware Recovery (DGX GB200, DGX GB300, OEM GB200 NVL72)

  • Autonomous Job Recovery (DGX GB200, OEM GB200 NVL72)

Planned for Future Release:

  • NetQ for DGX B200/B300 (shown as “NetQ soon”)

  • Autonomous Hardware Recovery for DGX B200/B300

  • Autonomous Job Recovery for DGX B200/B300 and DGX GB300

Not Supported:

  • Autonomous Job Recovery for OEM GB200 NVL72 (second variant shown in table)

Platform Comparison#

Table 1 Key Platform Differences#

Feature

B-Series (B200/B300)

DGX GB-Series (GB200/GB300)

OEM GB200 NVL72

System Type

Traditional DGX in standard racks

NVL72 rack-based with compute trays

NVL72 rack-based with compute trays

Architecture

x86 only

Hybrid ARM/x86

Hybrid ARM/x86

Control Planes

Single control plane

Separated admin/user planes (GB200)

Varies by configuration

NVLink Fabric

Not applicable

Yes (NVLink switches)

Yes (NVLink switches)

AHR/AJR Status

Planned

Available (v2.1.0 for GB300, v2.0.0 for GB200)

Varies by variant

Version Information#

Version:

2.1.0

Document Date:

December 2025

Vendor:

NVIDIA Corporation

Additional Resources#

For more information about NVIDIA Mission Control, please visit:

  • NVIDIA Base Command Manager Documentation

  • NVIDIA DGX Systems Documentation

  • NVIDIA GB200 NVL72 Documentation

  • NVIDIA Enterprise Support Portal

For hardware specifications and control plane requirements, refer to:

  • NVIDIA Mission Control Hardware Requirements Guide

  • DGX Product Architecture and Management Documentation