The NVIDIA Quantum-2-based MQM9790-NS2F switch comes with 64 400Gb/s ports on 32 physical octal small form-factor pluggable (OSFP) connectors that can be split to deliver up to 128 200Gb/s ports in a 1U standard chassis design. This switch use connector cages that house two 400Gb/s ports in a single cage called 2x400G (800G) twin-port OSFP and are used exclusively in these air-cooled switches. A single switch carries an aggregated bidirectional throughput of 51.2 terabits per second (Tb/s), with a landmark of more than 66.5 billion packets per second (BPPS) capacity.
Supporting the latest NVIDIA high-speed interconnect 400Gb/s technology, NVIDIA Quantum-2 brings a high-speed, extremely low-latency and scalable solution that incorporates state-of-the-art technologies such as Remote Direct Memory Access (RDMA), adaptive routing, and NVIDIA Scalable Hierarchical Aggregation and Reduction Protocol (SHARP)™.
Click to open expanded view
Product Highlights
MQM9790-NS2F, NVIDIA® 64-Port NDR 400G InfiniBand Data Center Switch, 32 OSFP Ports, Unmanaged, x86 Dual Core, NVIDIA Quantum™-2 Chip, P2C Airflow
Product Spotlights
Overcome link failures and achieves network recovery 5,000X faster than any software-based solution
Quickly resolve network issues due to specific port failures
SHIELD greatly reduces troubleshooting time
Support HDR, NDR and subsequent technologies
On-board Subnet Manager (SM) supporting fabrics of up to 2K nodes
Unified Fabric Manager (UFM) agent
Support NVIDIA Mellanox Scalable Hierarchical Aggregation and Reduction Protocol (SHARP)
Offload collective communication operations from the CPU to the switch network by aggregating large data vectors at wire rate
Accelerate HPC, deep learning algorithm performance
Lowest latency for Inter-GPU communication across nodes
Build networks with MLNX-OS
Enable scaling to thousands of compute and storage nodes
Provide monitoring and provisioning capabilities
Support CLI-based, using console ports
Support WEB UI, using 10/100/1000Mbps management port or service port
Support SNMP (SNMPv1, v2c, v3)
Support JSON
.svg)
Scalable Hierarchical Aggregation and Reduction Protocol (SHARP)™ technology offloads collective communication operations from the CPU to the switch network, crucial for computing applications.
With self-healing networking capabilities, it overcomes link failures and recoveries the network 5,000X faster than any software-based solution, achieving a stable and scalable system.
The adaptive routing algorithms help to prevent packet delivery failures. It also minimises network congestion and increases the network performance.
Ports | 64 x NDR 400G 32 OSFP Connectors | 64 x NDR 400G 32 OSFP Connectors |
Switching Capacity | 25.6 Tbps | 25.6 Tbps |
Power Supplies | 2 (1+1 Redundancy) Hot-swappable, AC | 2 (1+1 Redundancy) Hot-swappable, AC |
Fans | 7 (6+1 Redundancy) Hot-swappable, Back-to-Front | 7 (6+1 Redundancy) Hot-swappable, Back-to-Front |
Dimensions (Hx Wx D) | 1.7"x 17"x 26"(43.6x 438x 660mm) | 1.7"x 17"x 26"(43.6x 438x 660mm) |
Max. Power Consumption | 1593W | 1703w |
Management | Inband | Inband/outband |
Subnet Manager | ✓ | ✓ |
Adaptive routing | ✓ | ✓ |
VL2VL | ✓ | ✓ |
SHARP | ✓ | ✓ |
Qos | ✓ | ✓ |
Enhanced Qos | - | - |
Congestion control | ✓ | ✓ |
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓
✓