NVIDIA Switch-IB 2 SB7800 Series 100Gb/s InfiniBand Switch

1. Product Overview

The NVIDIA Switch-IB 2 SB7800 Series is a high-performance, 100Gb/s InfiniBand switch designed for modern data centers requiring extreme bandwidth, low latency, and scalable interconnect solutions. Ideal for HPC, AI, and large-scale cloud environments, this network switch delivers 36 non-blocking ports in a compact 1U form factor, offering an aggregate throughput of 7.2 Tb/s. It is engineered to reduce operational complexity while accelerating data-intensive workloads.

2. Key Characteristics

• 36 ports of 100Gb/s InfiniBand per port
• 7.2 Tb/s aggregate bidirectional throughput
• Ultra-low latency for high-performance applications
• Redundant, hot-swappable power supplies and fans
• Energy-efficient design with ENERGY STAR certification
• Comprehensive management via MLNX-OS and UFM platforms

3. Core Technology

The switch leverages NVIDIA’s Scalable Hierarchical Aggregation and Reduction Protocol (SHARP)™ technology, which offloads collective operations from CPUs to the switch network. This reduces MPI operation times and decreases network traffic. Support for advanced InfiniBand standards ensures seamless integration into high-performance fabric architectures.

4. Working Principle

The SB7800 operates as a top-of-rack leaf switch, routing data packets with minimal latency. Its in-network computing capability via SHARP allows aggregation and reduction operations to be performed within the switch, reducing the number of data transmissions required between nodes and freeing CPU resources for computation.

5. Applications

This network switch is ideal for:

• High-Performance Computing (HPC) clusters
• Artificial Intelligence and Machine Learning training networks
• Large-scale data analytics and cloud infrastructure
• Research and enterprise data centers

6. Specifications

7. Advantages

• Superior bandwidth and low latency for demanding applications
• In-network compute offload with SHARP technology
• High availability with redundant power and cooling
• Energy-efficient operation reduces TCO
• Flexible deployment in various data center topologies

8. Service & Support

We offer comprehensive support including warranty, 24/7 technical consultation, and integration services. Our team ensures timely delivery and post-sale support to maintain optimal performance of your network switch infrastructure.

9. FAQ

Q: What is the typical use case for the SB7800 switch?
A: It is designed for HPC, AI, and large data centers requiring high bandwidth and low latency.

Q: Does it support redundant power?
A: Yes, it comes with 1+1 redundant, hot-swappable power supplies.

Q: Can it be managed externally?
A: Yes, via NVIDIA UFM® platform for full fabric management.

10. Precautions

• Ensure proper airflow (P2C or C2P) based on model variant.
• Operate within specified temperature (0°C–45°C) and humidity ranges.
• Use compatible QSFP28 transceivers and cables.
• Verify EMC compliance (CE, FCC, VCCI) in your region.

11. Company Introduction

With over 10 years of industry experience, we operate a large factory supported by a strong technical team. We represent leading brands including Mellanox (now NVIDIA), Ruckus, Aruba, and Extreme. Our product portfolio includes original new network switches, network cards, wireless access points, controllers, and cables. We maintain a $10 million inventory to support bulk orders and provide 24/7 customer service and technical support. Our professional sales and engineering teams have earned a global reputation for reliability and quality service.