NVIDIA Mellanox 920-9B110-00FH-0D0 InfiniBand Switch Product Launch and Key Feature Overview

NVIDIA Mellanox 920-9B110-00FH-0D0 InfiniBand Switch｜Product Launch and Key Feature Overview

As HPC and enterprise data centers continue to demand proven 200Gb/s HDR interconnect, the NVIDIA Mellanox 920-9B110-00FH-0D0 offers a validated InfiniBand switching option for network engineers and architects. This ordering part number corresponds to the MQM8790-HS2F platform, positioned for top-of-rack leaf roles and mid-sized cluster deployments.

Background: Why This Product Matters

During the transition from EDR to HDR, many data centers do not need to jump immediately to 400Gb/s NDR, yet they still require higher port density and lower latency than EDR can provide. Legacy EDR switches are approaching their limits in per-port speed and aggregate bandwidth. The NVIDIA Mellanox 920-9B110-00FH-0D0 fills this generational gap with 40 QSFP56 ports running at 200Gb/s, delivering 16Tb/s of non-blocking switching capacity in a 1U form factor.

For budget-conscious projects that still require InfiniBand's deterministic low latency, the 920-9B110-00FH-0D0 InfiniBand switch OPN is a practical choice. It allows architects to build HDR fabrics without committing to the higher cost and cabling complexity of NDR, while preserving a migration path for future upgrades.

Key Technical Highlights

The 920-9B110-00FH-0D0 MQM8790-HS2F 200Gb/s HDR platform is designed around high radix and low latency. Its 40 QSFP56 ports support 200Gb/s HDR, and the switch is backward compatible with 100Gb/s EDR and 40Gb/s FDR links, which simplifies mixed-generation deployments. Key characteristics include:

• 40 x 200Gb/s QSFP56 ports with 16Tb/s aggregate non-blocking bandwidth
• Support for RDMA, MPI, and NVIDIA SHARP for collective operation offload
• Advanced congestion control and adaptive routing for lossless transport
• Compatibility with standard InfiniBand management and telemetry tools
• 1U form factor suitable for top-of-rack and leaf deployments

For teams evaluating 920-9B110-00FH-0D0 specifications, the switch is part of the NVIDIA Quantum family and interoperates with existing HDR adapters and cables. Reviewing the 920-9B110-00FH-0D0 datasheet early helps confirm power, cooling, and rack requirements. Architects should also verify 920-9B110-00FH-0D0 compatible host channel adapters and transceivers before finalizing a bill of materials.

Applications and Deployment Scenarios

The 920-9B110-00FH-0D0 InfiniBand switch OPN solution is well suited to HPC clusters, AI inference farms, and storage fabrics that need predictable latency at 200Gb/s per link. In a two-tier fabric, these switches typically serve as leaf devices connecting compute nodes and storage targets, while spine capacity can be added as the cluster grows.

Procurement teams often ask about 920-9B110-00FH-0D0 price and 920-9B110-00FH-0D0 for sale channels. Pricing varies by configuration, support level, and region, so it is best to consult authorized distributors and NVIDIA enterprise representatives for current quotes. 

Summary

As a 200Gb/s HDR InfiniBand switch with 40 QSFP56 ports, the NVIDIA Mellanox 920-9B110-00FH-0D0 delivers the density, low latency, and backward compatibility needed for pragmatic HDR deployments. For network engineers and IT managers planning a fabric refresh, it represents a cost-effective and standards-aligned building block for HPC and enterprise InfiniBand networks.