NVIDIA Mellanox MQM9790-NS2F InfiniBand Switch – Product Launch & Key Feature Deep-Dive

July 13, 2026

ultime notizie sull'azienda NVIDIA Mellanox MQM9790-NS2F InfiniBand Switch – Product Launch & Key Feature Deep-Dive

NVIDIA Mellanox MQM9790-NS2F InfiniBand Switch – Product Launch & Key Feature Deep-Dive

The data center landscape is shifting rapidly, with AI workloads, high-performance computing (HPC), and large-scale distributed storage demanding ever-higher bandwidth and lower latency. Traditional Ethernet fabrics often struggle to deliver the deterministic performance and congestion control required for these mission-critical applications. Enter the NVIDIA Mellanox MQM9790-NS2F – a next‑generation InfiniBand switch that redefines the backbone of modern computing clusters.

Background: Why a Dedicated InfiniBand Switch?

As GPU‑based AI training and simulation workloads scale to thousands of nodes, network throughput and latency become the primary bottlenecks. The MQM9790-NS2F InfiniBand switch addresses this by offering a purpose‑built fabric that eliminates packet loss, provides sub‑microsecond latency, and delivers massive aggregate bandwidth. For network engineers and architects, this means predictable job completion times and higher cluster efficiency – a critical advantage over shared Ethernet infrastructures.

Key Technical Capabilities

At the heart of the NVIDIA Mellanox MQM9790-NS2F lies the latest NDR (Next Data Rate) technology. This switch offers 64 OSFP ports, each capable of 400 Gb/s full‑duplex bandwidth, resulting in an aggregate switching capacity of over 25 Tb/s. The MQM9790-NS2F 400Gb/s NDR 64-port OSFP configuration enables ultra‑dense top‑of‑rack or leaf‑spine deployments, significantly reducing cable complexity while maximizing port density per rack unit.

  • Cut‑through switching & adaptive routing – Dynamically reroutes traffic to avoid congestion, ensuring consistent low latency even under heavy load.
  • SHARPv3 (Scalable Hierarchical Aggregation and Reduction Protocol) – Offloads collective communication operations, accelerating AI training by up to 20% compared to software‑based approaches.
  • In‑network computing – Supports advanced data reduction and segmentation, reducing host CPU overhead and improving overall application performance.

For those evaluating the MQM9790-NS2F datasheet, key specifications include a 1U form factor, redundant hot‑swappable power supplies, and support for both copper and optical OSFP modules. The MQM9790-NS2F specifications also highlight advanced telemetry and monitoring capabilities, enabling IT managers to gain real‑time visibility into fabric health and traffic patterns – essential for proactive maintenance and capacity planning.

Deployment Scenarios & Compatibility

The MQM9790-NS2F InfiniBand switch solution is designed to integrate seamlessly with NVIDIA’s Quantum‑2 family, ensuring MQM9790-NS2F compatible operation with existing ConnectX‑7 and BlueField‑3 adapters. Whether you are building a new AI supercomputer or upgrading an existing HPC cluster, this switch delivers the performance and scalability required for exascale‑class workloads. Common use cases include:

  • Large‑scale AI training clusters (GPT‑style models, computer vision)
  • Weather simulation and computational fluid dynamics
  • Distributed database and analytics platforms

For procurement teams, the MQM9790-NS2F price is competitive when measured against the total cost of ownership – lower power consumption per port and reduced cabling costs translate to significant savings over a 5‑year lifecycle. Early adopters can already check MQM9790-NS2F for sale through authorized NVIDIA channel partners, with general availability expected in Q3 2026.

Summary: A New Benchmark for InfiniBand Fabrics

The NVIDIA Mellanox MQM9790-NS2F is not merely an incremental upgrade – it is a fundamental leap in switch architecture, combining massive port density, 400 Gb/s NDR speed, and intelligent in‑network computing. For network architects and IT managers, it offers a future‑proof foundation that can scale from dozens to thousands of nodes without compromising performance. With its robust feature set and seamless compatibility, the MQM9790-NS2F InfiniBand switch is poised to become the go‑to solution for the next generation of data‑intensive computing.

To dive deeper into the MQM9790-NS2F datasheet or explore reference designs, visit the NVIDIA Networking portal or contact your local sales representative. The era of intelligent, high‑speed fabrics is here – and it starts with the MQM9790-NS2F.

Feature MQM9790-NS2F Specification
Ports & Speed 64 OSFP, 400 Gb/s NDR per port
Switching Capacity 25.6 Tb/s aggregate
Latency < 600 ns (cut‑through)
Power Efficiency ~1.5 W per port
Management NVIDIA UFM, SNMP, Redfish