MFS1S90-H0xxE 2x200Gb/s QSFP56 to 2x200Gb/s QSFP56 MMF AOC Product Specifications

Introduction

NVIDIA® MFS1S90 is a QSFP56 VCSEL-based (Vertical Cavity Surface-Emitting Laser), cost-effective 2x 200Gb/s to 2x 200Gb/s active optical splitter cable (AOC), designed for use in 200Gb/s InfiniBand (IB) HDR (High Data Rate) systems. It provides cross-connect capability between ToR (Top of Rack) and Spine switches. The cable enables a HDR InfiniBand QSFP56 switch port to operate as 2x HDR100. It creates a non-blocking fat-tree topology doubling the number of ports. This eliminates a third level of switching, which is needed for traditional cluster design of 1600 up to 3200 HDR100 ports, and results in significant CAPEX and OPEX savings. By reducing a layer of switching the cluster latency is reduced, which results in improved performance.

AOC Splitter
H-Cable

Up to 3200 Ports
2-Level Fat-Tree

image2021-8-22_13-29-58.png

image2021-8-22_13-30-14.png

Each QSFP56 end of the cable comprises an EEPROM providing product and status monitoring information, which can be read by the host system.

NVIDIA’s unique-quality cable solutions provide power-efficient connectivity for data center interconnects. It enables higher port bandwidth, density and configurability at a low cost, and reduced power requirement in the data centers. Rigorous production testing ensures the best out-of-the-box installation experience, performance and durability.

image2021-8-22_13-32-10.png
Warning

Images are for illustration purposes only. Product labels, colors, and lengths may vary.

Key Features

  • 2x 200Gb/s (2x 100Gb/s HDR100) to 2x 200Gb/s (2x 100Gb/s HDR100)

  • 4x 50Gb/s PAM4 modulation

  • Programmable Rx output amplitude and pre-emphasis

  • SFF-8665 compliant QSFP56 port

  • Single 3.3V power supply

  • 4.5W power consumption (typ., each end)

  • Up to 30m length

  • Hot pluggable

  • RoHS compliant

  • SFF-8636 compliant I2C management interface

Application

  • Connects between 200G ToR switches and Spine switches with fat-tree topology

© Copyright 2023, NVIDIA. Last updated on May 22, 2023.