Friday, September 27, 2024
10:20 – 11:10 a.m. (CST)
ETB 1020
Dr. Pallab Datta
Senior Research Scientist, Brain-Inspired Computing
IBM Research – California
Title: “Neural Inference at the Frontier of Energy, Space, and Time”
Abstract
Computing, since its inception, has been processor-centric, with memory separated from compute. Inspired by the organic brain and optimized for inorganic silicon, NorthPole is a neural inference architecture that blurs this boundary by eliminating off-chip memory, intertwining compute with memory on-chip, and appearing externally as an active memory chip. NorthPole is a low-precision, massively parallel, densely interconnected, energy-efficient, and spatial computing architecture with a co-optimized, high-utilization programming model. On the ResNet50 benchmark image classification network, relative to a graphics processing unit (GPU) that uses a comparable 12-nanometer technology process, NorthPole achieves a 25 times higher energy metric of frames per second (FPS) per watt, a 5 times higher space metric of FPS per transistor, and a 22 times lower time metric of latency. Similar results are reported for the Yolo-v4 detection network. NorthPole outperforms all prevalent architectures, even those that use more-advanced technology processes. Also, some new results will be presented on LLMs and a 3U VPX board.
Biography
Pallab Datta is a member of the Brain-Inspired Computing group at IBM Almaden Research Center, which developed the NorthPole Neural Inference processor and the TrueNorth neurosynaptic processor. He is currently a senior research scientist and technical lead for Compiler development for NorthPole. He joined IBM in 2010 to work on the DARPA funded IBM SyNAPSE Project. As part of the TrueNorth project he was involved in the development of the Corelet Programming Language for programming the reconfigurable neuromorphic hardware. He was also involved in the development of algorithms and applications with networks of neurosynaptic cores for building cognitive systems. He had also worked on large-scale simulations using the IBM Neuro Synaptic Core Simulator (Compass) as part of the IBM SyNAPSE Project.
Before joining IBM Research, he had worked at The NeuroSciences Institute and Los Alamos National Laboratory. He was also a visiting researcher at INRIA Sophia-Antipolis . He completed his PhD at Iowa State University in 2005 under the supervision of Prof. Arun K. Somani.
In general, his technical interests include machine learning, AI, Compilers, AI hardware architecture and simulations, High-performance & distributed computing & optimization techniques.
He has authored in several international journals and conferences, and is a member of the Institute of Electrical and Electronics Engineers (IEEE) and ACM.
To learn more about Dr. Datta go to https://research.ibm.com/people/pallab-datta.
You are welcome to join us on Friday, 9/27/24 at 10:20 a.m. in ETB 1020!