NVIDIA SWITCH-IB 2 SB7800 SERIES
Accelerating Data Centers With InfiniBand Switches
36-port Non-Blocking Externally-Managed EDR 100Gb/s InfiniBand Smart Switch
non-PoE, AFI/P2C (Back-to-front/Connector-side-outlet airflow)
Benefits
• Industry-leading switch platform in performance, power, and density
• Designed for energy and cost savings
• Collective communication acceleration
• Quick and easy setup and management
• Maximizes performance by removing fabric congestions
• Backward compatible to FDR technology
NVIDIA Networking provides the world's first smart switch, enabling in-network computing through the Co-Design Scalable Hierarchical Aggregation and Reduction Protocol (SHARP) technology. SB7890 has the highest fabric performance available in the market with up to 7.2Tb/s of non-blocking bandwidth with 90ns port-to-port latency.
Faster servers based on PCIe 3.0, combined with high-performance storage and applications that use increasingly complex computations, are causing data bandwidth requirements to spiral upward. As servers are deployed with next generation processors, High-Performance Computing (HPC) environments and Enterprise Data Centers (EDC) will need every last bit of bandwidth delivered with next generation of Extended Data Rate (EDR) InfiniBand high-speed smart switches.
Features
Performance
• 36 x EDR 100Gb/s ports in a 1U switch6 x EDR 100Gb/s ports in a 1U switch
• 7.2Tb/s aggregate switch throughput
• 90ns switch latency
• 122W typical power consumption
Optimized design
• 1+1 redundant & hot-swappable power
• N+1 redundant & hot-swappable fans
• 80 gold+ and energy star certified power supplies
Advanced design
• Adaptive routing
• Congestion control
• VL2VL mapping
• Collective offloads (SHARP technology)
Resilience and Ease of Scale
Customers seeking to take advantage of the vast amount of data generated daily by high performance computing (HPC) and AI technologies, are increasingly finding themselves needing every bit of the bandwidth delivered by NVIDIA® InfiniBand high-speed smart switches.
In-Network Acceleration
High performance computing workloads require intelligent interconnect solutions to achieve the highest possible application performance. NVIDIA Scalable Hierarchical Aggregation and Reduction Protocol (SHARP)™ technology improves upon the performance of MPI operations, by offloading collective operations from the CPU to the switch network thus eliminating the need to send data multiple times between endpoints. This innovative approach decreases the amount of data traversing the network as aggregation nodes are reached, and dramatically reduces the MPI operations time. Additional benefits like collective communication, free up valuable CPU resources for computation rather than using them to process communication.
Reduced IT Costs and Complexity
NVIDIA SB7800 switches deliver 36 ports of 100 gigabits per second (Gb/s) InfiniBand per port in a standard 1U standard chassis design. A single switch carries an aggregated bidirectional throughput of 7.2 terabits per second (Tb/s) and ultra-low port latency, both improving application performance and scalability, and reducing operational costs and IT infrastructure complexity. As an ideal top-of-rack leaf connectivity solution for small to extremely large clusters, the internally managed SB7800 and the externally managed SB7890 100Gb/s InfiniBand fixed-configuration switches allow flexibility through enabling a variety of topologies.
World-Class Design
SB7800 switches are designed for performance, serviceability, energy savings and high-availability. The power-optimized design enables these switches to support low power consumption with ATIS weighted power consumption as low as 122W for a fully populated system. This means more power reduction if not all ports are fully utilized. The SB7800 also has redundant power supplies (1+1) and fans that boast achieving maximal thermal protection.
Enhanced Management
The SB7800’s dual-core x86 CPU runs MLNX-OS® software packages, which deliver complete chassis management of the switch’s firmware, power supplies, fans and ports. The switch can also be coupled with NVIDIA UFM® platforms, with their comprehensive feature set to meet the broadest range of modern, scale-out InfiniBand data center requirements. Revolutionizing data center management, the UFM family of products combines enhanced, real-time network telemetry with AI-powered cyber intelligence and analytics. The UFM solutions minimize cluster downtime by empowering research and industrial data center operators to efficiently provision, monitor, manage, and preventatively troubleshoot and maintain their InfiniBand data center fabric.
Full specification and details can be found in the Product Datasheet PDF file