Kunle Olukotun, Cadence Design Professor of Electrical Engineering and Computer Science at Stanford University, will give a presentation during the CAII Seminar Series on Monday, April 26 at 11:00 a.m. The talk is titled “Dataflow Optimized Systems for AI."
View Seminar here: https://go.ncsa.illinois.edu/CAIISpringSemesterSeriesSP21
In many applications, traditional software development is being replaced by machine learning generated models resulting in accuracy improvements and deployment advantages. This fundamental shift in how we develop software, referred to as Software 2.0, has provided dramatic improvements in application quality and ease of deployment. Additionally, these developments are enabling a new architectural approach where the applications “program” the hardware to provide optimal dataflow, efficiency and performance.
In this talk I will describe the full-stack design approach we are taking at SambaNova to exploit the characteristics of AI applications to maximize the capabilities of dataflow optimized hardware.The SambaFlow compiler stack transparently compiles models expressed in Machine Learning frameworks to optimized dataflow pipelines that exploit task, data, and pipeline parallelism. SambaFlow unlocks new dimensions of performance and accuracy on a range of AI applications such as hi-resolution images, recommender models with large embeddings, and natural language processing models with huge parameter counts.
Kunle Olukotun is the Cadence Design Professor of Electrical Engineering and Computer Science at Stanford University. Olukotun is well known as a pioneer in multicore processor design and the leader of the Stanford Hydra chip multiprocessor (CMP) research project.
Prior to SambaNova Systems, Olukotun founded Afara Websystems to develop highthroughput, low-power multi-core processors for server systems. The Afara multi-core processor, called Niagara, was acquired by Sun Microsystems. Niagara derived processors now power Oracle’s SPARC-based servers.
Olukotun is the Director of the Pervasive Parallel Lab and a member of the Data Analytics for What’s Next (DAWN) Lab which is developing infrastructure for usable machine learning.
Olukotun is an ACM Fellow and IEEE Fellow for contributions to multiprocessors on a chip and multi-threaded processor design. Olukotun recently won the prestigious IEEE Computer Society’s Harry H. Goode Memorial Award and was also elected to the National Academy of Engineering, which is one of the highest professional distinctions accorded engineers.
Kunle received his Ph.D. in Computer Engineering from The University of Michigan.
This presentation will be recorded and will be available on the CAII website shortly after the presentation.