FS-MQM8700-HS2F, NVIDIA® 40-Port 200G InfiniBand Data Center Switch, 40 x HDR FS-QSFP56, Managed, x86 Dual Core, NVIDIA Quantum™ Chip, P2C Airflow

The NVIDIA Quantum-based FS-MQM8700-HS2F switch comes with 40 200Gb/s ports on 40 physical Quad Small Form-factor Pluggable 56 (QSFP56) connectors that can be split to deliver up to 80 100Gb/s ports in a 1U standard chassis design. A single switch carries an aggregated bidirectional throughput of 16 terabits per second (Tb/s), with sub 130ns port-to-port latency.

Supporting the latest NVIDIA high-speed interconnect 200Gb/s technology, NVIDIA Quantum 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)™.

Specifications

• Manufacturer: Mellanox
  
• Ports: 40 x HDR 200G; 40 QSFP56 Connectors
  
• CPU: Broadwell ComEx D-1508 2.2GHZ
  
• Switch Chip: NVIDIA QUANTUM
  
• Software: MLNX-OS
  
• Switching Capacity: 8 Tbps
  
• Typical Power Consumption: 253W
  
• System Memory: Single 8GB
  
• Max. Power Consumption: 784W
  
• Latency: 130ns
  
• Voltage: 100-127VAC 50/60Hz 4.5A; 200-240VAC 50/60Hz 2.9A
  
• AC Power Supplies: 1+1 Hot-swappable
  
• Fan Number: 5+1 Hot-swappable
  
• Airflow: Back-to-Front (P2C)
  
• Operating Temperature: 32°F to 104°F (0ºC to 40ºC)
  
• Storage Temperature: -40°F to 158°F (-40ºC to 70ºC)
  
• Dimensions (Hx Wx D): 1.7"x 16.85"x 23.2"(43.6x 433.2x 590.6mm)
  
• Rack Units: 1 RU
• Ports: 40 x HDR 200G 40 QSFP56 Connectors
• Switching Capacity: 8 Tbps
• Power Supplies: 2 (1+1 Redundancy) Hot-swappable, AC
• Fans: 6 (5+1 Redundancy) Hot-swappable, Back-to-Front

Product Spotlights

High Availability And Multi-Service Support

• 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 EDR, HDR and subsequent technologies 
• On-board Subnet Manager (SM) supporting fabrics of up to 2K nodes 
• Unified Fabric Manager (UFM) agent
• Max. Power Consumption: 784W
• Management: Inband/outband
• Subnet Manager: YES
• Adaptive routing: YES
• VL2VL: YES
• SHARP: YES
• Qos: YES
• Enhanced Qos: YES
• Congestion control: YES

Error-Free Network Configuration

• 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

Secure And Simplified Access for Users And IoT

• 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 ports
• Support SNMP (SNMPv1,v2c,v3)
• Support JSON

Connectivity Solutions

Quantum InfiniBand Switches Deliver the Highest Performance and Port Density

Non-blocking or Bidirectional Bandwidth Ensures Flexible Network
FS-MQM8700-HS2F provides up to forty 200Gb/s ports with non-blocking bandwidth or eighty 100Gb/s ports with full bidirectional bandwidth, achieving a low latency and flexible network.

Hot-swappable Redundant Power and Fans
Hardware-level redundancy protection to ensure business continuity. 80 gold+ and energy star certified powers save money on your power bill, and improves longevity and reliability.

In-Network Computing Enable Higher Performance and Scale
Scalable Hierarchical Aggregation and Reduction Protocol (SHARP)™ technology offloads collective communication operations from the CPU to the switch network, crucial for computing applications.

Self-Healing Ability Enhances System Performance and Utilization
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.

InfiniBand Proven Adaptive Routing Performance
The adaptive routing algorithms help to prevent packet delivery failures. It also minimises network congestion and increases the network performance.

Multiple Management Methods for Easy Network Maintenance
Supporting SNMP, CLI, WEB, JSON, and advanced NVIDIA Unified Fabric Manager (UFM®) platform to realize higher utilization and reduce overall opex of the modern data center network.