Introduction

The NVIDIA Quantum-2-based QM9700 and QM9790 switch systems deliver an unprecedented 64 ports of NDR 400Gb/s InfiniBand per port in a 1U standard chassis design. 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 NDR 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)™.Unlike any other networking solution, NVIDIA InfiniBand provides self-healing network capabilities, as well as quality of service (QoS), enhanced virtual lane (VL) mapping, and congestion control to provide the highest overall application throughput. As an ideal rack-mounted InfiniBand solution, the QM9700 and QM9790 NDR InfiniBand fixed-configuration switches allow maximum flexibility, as they enable a variety of topologies, including Fat Tree, SlimFly, DragonFly+, multi-dimensional Torus, and more. They’re also backwards compatible to previous generations and include expansive software ecosystem support.

Today’s complex research demands ultra-fast processing of high-resolution simulations, extreme-size datasets, and complex, highly parallelized algorithms that need to exchange information in real time. The QM9700 NDR InfiniBand switches extend NVIDIA In-Network Computing technologies and introduce the third generation of NVIDIA SHARP technology, SHARPv3. Creating virtually unlimited scalability for large data aggregation through the data center network, participating in the application’s runtime and reducing the amount of data needed to traverse the network.

By implementing NVIDIA port-split technology, the QM9700 and QM9790 switches provide a double-density radix for 200Gb/s (NDR200) data speeds, reducing the cost of network design and network topologies. Supporting up to 128 ports of 200Gb/s, NVIDIA delivers the densest top-of-rack (TOR) switch available on the market. The QM9700 family of switches enables small to medium-sized deployments to scale with a two-level Fat Tree topology while reducing power, latency, and space requirements.

The internally managed QM9700 switch features an on-board subnet manager that enables simple, out-of-the-box bringup for up to 2,000 nodes. Running the NVIDIA MLNX-OS® software package, the subnet manager delivers full chassis management through command-line interface (CLI), web-based user (WebUI), Simple Network Management Protocol (SNMP), or JavaScript Object Notation (JSON) interfaces.The externally managed QM9790 switch can utilize the advanced NVIDIA Unified Fabric Manager (UFM®) feature sets to empower data center operators to efficiently provision, monitor, manage, preventatively troubleshoot, and maintain the modern data center fabric, to realize higher utilization and reduce overall opex.

QM9700 Front View

QM9790 Front View

QM9700 and QM9790 Rear View

For additional airflow options, see Airflow.

The table below describes maximum throughput and interface speed per system model.

64 NDR non-blocking ports with aggregate data throughput up to 51.2Tb/s

The table below lists the various management interfaces and available replacement parts per system model.

For a full feature list, please refer to the system’s product brief. Go to https://www.nvidia.com/en-us/networking/. In the main menu, click on Products > InfiniBand > Switch Systems , and select the desired product page.

The list of certifications (such as EMC, Safety and others) per system for different regions of the world is located on the Mellanox website at http://www.mellanox.com/page/environmental_compliance.