Carnegie Mellon University

Webinar - Technical Overview of the Cerebras CS-1, the AI Compute Engine for Neocortex

Presented on Wednesday, August 19, 2020, 2:00 - 3:00 pm (ET), by Natalia Vassilieva, Ph.D. (Cerebras Systems Inc.)

In this webinar, we offer a technical deep dive into the Cerebras CS-1 system, the Wafer Scale Engine (WSE) chip, and the Cerebras Graph Compiler. The Cerebras CS-1 is the groundbreaking specialized AI hardware technology to be featured on Neocortex, PSC’s upcoming NSF-funded AI supercomputer. Neocortex, planned for deployment in late 2020, will enable on-chip model and data parallelism and will significantly accelerate the training of ambitious deep learning.

Access the recording, slides, and other content related to this webinar

Abstract

The most significant constraint on deep learning research and AI innovation today is long training times. Training deep neural network models with large real-world datasets commonly takes days, weeks, or even months, using computing infrastructure that was fundamentally designed for other work. To speed up training, researchers today must use large clusters of processors and distributed scale-out implementations of their training jobs. While this does reduce wall-clock training time for a single run, it is inefficient, requires time and expertise in distributed programming, and entails extensive model and hyperparameter tuning that impedes research. In contrast, the Cerebras CS-1 – powered by its massive, 400,000 core domain-specific Wafer-Scale Engine processor – provides the AI compute resources of a cluster with the usability and ease of use of a single device. In addition to that, the CS-1 is uniquely advantaged for novel techniques and emergent model architectures that challenge traditional AI accelerators. 

This webinar includes:
  • Technical overview of the Cerebras Wafer Scale Engine (WSE), the world’s only trillion-transistor processor;
  • Introduction to the CS-1 system, hosting the WSE, and built to power, cool and bring data to the WSE;
  • And a technical intro to the Cerebras Graph Compiler, tightly co-designed with the WSE to be able to take full advantage of its computational resources while allowing researchers to program using popular deep learning frameworks such as TensorFlow.
Access the recording, slides, and other content related to this webinar

About the Instructor

Natalia Vassilieva, Ph.D., (Cerebras Systems Inc. | Computer Scientist )

Dr. Vassilieva is a Sr. Technical Product Manager at Cerebras Systems, an innovative computer systems company dedicated to accelerating deep learning. Natalia’s main interests and expertise are in machine learning, artificial intelligence, analytics, and application-driven software-hardware optimization and co-design. Prior to Cerebras, Dr. Vassilieva was affiliated to Hewlett Packard Labs where she led the Software and AI group from 2015 till 2019 and served as the head of HP Labs Russia from 2011 to 2015. From 2012 to 2015, Natalia also served as a part-time Associate Professor at St. Petersburg State University and a part-time lecturer at the Computer Science Center, St. Petersburg, Russia. Before joining HP Labs in 2007, Natalia worked as a Software Engineer for different IT companies in Russia from 1999 till 2007. Natalia holds a Ph.D. in Computer Science from St. Petersburg State University.

For more information about Neocortex, explore the project page. For questions about this webinar, please email neocortex@psc.edu.