Research Projects

Fast Network Interface Cards (FastNICs)

DARPA FastNICs program targets speeding up applications such as the distributed training of machine learning classifiers by 100x through the development, implementation, integration, and validation of novel, clean-slate network subsystems. The program focuses on overcoming the gross mismatches in computing and network subsystem performance. Columbia leads the key hardware innovation - Optical Network Interface Cards (O-NICs) in the FLEET project to meet or exceed the FastNICs program goals.

NSA

This project focuses on development of high performance computing related models and network simulations, and to expand our in-house simulation engine, PhoenixSim, to support the automation and configuration of SST simulation scripts with graphic user interface.  

Optically Connected Memory (OCM)

Computing platforms are increasingly challenged by the need to process growing volumes of data arriving in real time and at very high rates. New memory technologies such as Hybrid Memory Cube (HMC) and High-bandwidth Memory (HBM) provide a Tbps memory access. Optical interconnects based on Silicon Photonic platform enable new distance independent, and reconfigurable compute to memory architectures. Our work spans from the physical layer including transceivers and switches to application demonstration of memory bound applications.

Photonic Integrated Networked Energy efficient datacenter (PINE)

The Photonic Integrated Networked Energy efficient datacenter (PINE)vision is built on three main innovation pillars: i) redesigning the architecture based on disaggregation of compute and memory resources over a unified photonic interconnect with bandwidth steering capabilities to improve resource allocation throughout the datacenter, ii) introducing a concept of embedded datacenter nodes consisting of various Multi-Chip Modules (MCMs) interconnected in a unique interposer platform via high bandwidth density integrated photonics, and iii) developing a new generat

Photonic-Storage Subsystem Input/Output (P-SSIO) Interface

The Photonic-Storage Subsystem Input/Output (P-SSIO) Interface project aims to provide low-power high I/O bandwidth capacity for the storage subsystem in leadership superclass computers. Energy -efficient Photonic transceiver and Photonic-switched interconnect enable reconfigurable connectivity between hosts and storage devices for the disaggregated system architectures.

Photonics in the Package for Extreme Scalability (PIPES)

Computing systems are critically challenged to meet the performance of data intensive applications across multiple platforms particularly driven by the explosive growth in data analytics. Data movement, dominated by energy costs and limited ‘chip-escape’ bandwidth densities, is perhaps the singular roadblock to these systems’ scalability. Integrated silicon photonics with deeply embedded optical connectivity is on the cusp of enabling revolutionary data movement and extreme performance capabilities.

Space-Based Adaptive Communications Node (Space-BACN)

The Space-BACN program aims to revolutionize the way space-based communications work by developing low-cost, high-speed reconfigurable optical datalinks to connect various low-earth orbit (LEO) constellations. This reconfigurable transceiver photonic integrated circuit (PIC) will be able to support intensity and phase modulation formats, over a wide range of data rates, in order to act as a bridge for communication between different LEO satellite constellations that each operate on their own unique data protocols.

 

Contact Us

Lightwave Research Labratory, Columbia University

530 W 120th St,

8th floor, Schapiro (CEPSR) Building

New York City, NY 10027