The Emu Migratory Thread Architecture

Departments: Electrical Engineering
Time: Thursday, October 31, 2019 - 4:00pm - 5:00pm
Type: Seminar Series
Presenter: Marty Deneroff; Chief Operating Officer, Emu Technology
Room/Office: Room 200
Location:
Arthur K. Watson Hall
51 Prospect Street
New Haven, CT 06511
United States
See map: Google Maps

Department of Electrical Engineering Seminar

“The Emu Migratory Thread Architecture”

Marty Deneroff
Chief Operating Officer
Emu Technology

Abstract: Emu Technology has developed a new computer architecture specifically aimed at efficient operation on large graphs and sparse matrices. It is based on the premise that, for big / sparse data, it is preferable to move the site of program execution to the locale of the data, rather than moving the data to the node where the program thread is executing. This talk will explain the motivations for this choice, describe the computer architecture that Emu has developed to implement the premise, and discuss the approaches and tradeoffs used in the implementation.

Bio: Marty Deneroff is the Chief Operating Officer of Emu Technology, Inc. Marty has spent most of his career in advanced computer architecture, particularly in the area of shared memory multiprocessors. His experience includes VP Server and Micro-Processor Engineering at SGI where he was responsible for delivering SGI's Distributed Shared Memory computers, and VP Engineering at D.E. Shaw Research where he was responsible for the development of the Anton molecular dynamics simulation engine. Marty has 30+ yrs. experience in the computer industry in a variety of capacities, and holds over twenty patents in various aspects of multiprocessor systems. He holds a BSEE from MIT and an MSCS from Monmouth University.

Hosted by: Professor Jakub Szefer & Richard Lethin

Thursday, October 31, 2019
Arthur K. Watson Hall, Room 200
4:00 PM