Brand: NVIDIA | Category: Network Switches
SKU: QM9800 | Part #: QM9800 | MPN: QM9800
Contact for Pricing — Request a Quote
The NVIDIA Quantum-3 InfiniBand QM9800 delivers 64 ports of 400Gb/s NDR InfiniBand connectivity with 51.2Tb/s aggregate switching capacity in a compact 1U form factor. This next-generation switch provides exceptional port density for hyperscale data center deployments, enterprise AI clusters, and high-performance computing (HPC) environments requiring ultra-low latency and massive bandwidth aggregation. Network engineers and AI infrastructure teams select the QM9800 to eliminate fabric bottlenecks while maintaining sub-100ns cut-through latency across the entire switching architecture.
NVIDIA engineered the Quantum-3 series with intelligent in-network computing via SHARP v3, enabling real-time collective operations that offload compute from CPU cores. Hardware-based adaptive routing and enhanced congestion control optimize traffic patterns dynamically, while backward compatibility with HDR 200Gb/s endpoints protects existing infrastructure investments. The QM9800 supports multiple topologies—fat-tree, dragonfly, dragonfly+, and hypercube—giving data center architects flexibility in cluster design. NVIDIA Onyx (MLNX-OS) and NVIDIA Unified Fabric Manager (UFM) streamline provisioning, monitoring, and multi-switch orchestration across enterprise fabrics. Redundant hot-swappable power supplies and flexible airflow options ensure operational resilience in demanding rack environments. OSFP transceiver support accommodates AOC and DAC interconnects, simplifying cabling strategies for high-density deployments.
Contact Omnixon Global in Dubai to request an RFQ for the NVIDIA Quantum-3 InfiniBand QM9800 and receive expert guidance on fabric integration and deployment planning.
| Brand | NVIDIA |
| Category | Network Switches |
| SKU | QM9800 |
| Part Number | QM9800 |
| Condition | New |
| Capacity | 800GB |
| Speed | 800Gb/s |
| Manufacturer Part Number | QM9800 |
| Switch Series | Quantum-3 |
| InfiniBand Standard | NDR (Next Data Rate) InfiniBand |
| Number of Ports | 64 |
| Per-Port Bandwidth | 400Gb/s NDR |
| Aggregate Switching Capacity | 51.2Tb/s |
| Form Factor | 1U rack-mountable |
| Latency | Sub-100ns cut-through latency |
| In-Network Computing | NVIDIA SHARP v3 (Scalable Hierarchical Aggregation and Reduction Protocol) |
| Adaptive Routing | Hardware-based adaptive routing supported |
| Congestion Control | Enhanced hardware-based congestion control |
| Backward Compatibility | HDR 200Gb/s InfiniBand endpoints supported |
| Management Interface | NVIDIA Unified Fabric Manager (UFM) compatible; out-of-band Ethernet management port |
| Supported Topologies | Fat-tree, dragonfly, dragonfly+, hypercube |
| Cable Support | Active Optical Cables (AOC), Direct Attach Copper (DAC), OSFP transceivers |
| Connector Type | OSFP (Octal Small Form Factor Pluggable) |
| Power Supply | Redundant hot-swappable power supplies |
| Cooling | Front-to-rear and rear-to-front airflow options |
| Operating System | NVIDIA Onyx (MLNX-OS) |
The NVIDIA Quantum-3 InfiniBand - 800Gb/s (QM9800) is enterprise networking hardware for data-center, campus, and branch deployments. Common roles include core/distribution switching, server uplinks, top-of-rack fabric, and edge connectivity in mixed Cisco/Juniper/Arista environments.
Key specifications for the NVIDIA Quantum-3 InfiniBand - 800Gb/s (QM9800): new condition; network type InfiniBand Switch; network port config 32x 800Gb/s NDR; network speed 800Gb/s; network switching capacity 51.2 Tbps; network forwarding rate 38.1 Bpps; network power consumption 800W. Manufacturer part number QM9800. For the full datasheet with electrical, environmental, and compliance details, contact our pre-sales engineering team.
The NVIDIA Quantum-3 InfiniBand - 800Gb/s (QM9800) interoperates with standard switching protocols (LACP, MLAG/VPC, 802.1Q, OSPF/BGP). For specific cross-vendor scenarios — Cisco-to-Arista MLAG peering, fabric integration with Cumulus/SONiC, or migration from legacy Catalyst — our network engineers map a compatibility plan as part of the RFQ.