Stanford Seminar - Computational memory: A stepping-stone to non-von Neumann computing?

EE380: Computer Systems Colloquium Seminar
Computational memory: A stepping-stone to non-von Neumann computing?
Speaker: Abu Sebastian, IBM Research - Zürich

In the advent of the data-centric AI era and the imminent end of CMOS scaling laws, the time is ripe to adopt computing units based on non-von Neumann computing architectures. A first step in this direction could be in-memory computing, where certain computational tasks are performed in place in a specialized memory unit called computational memory. Resistive memory devices, where information is represented in terms of atomic arrangements within tiny volumes of material, are poised to play a key role as elements of such computational memory units. I will present a few examples of how the physical attributes and dynamics of these devices can be exploited to achieve in-place computation. We expect that this co-existence of computation and storage at the nanometer scale could enable ultra-dense, low-power, and massively-parallel computing systems.

About the Speaker:
Abu Sebastian is a Research Staff Member and Master Inventor at IBM Research - Zürich. He was a contributor to several key projects in the field of storage and memory technologies. Most recently, he has been pursuing research in the area of non-von Neumann computing with the intent of connecting the technological elements with applications such as machine learning. In 2015, he was awarded a European Research Council (ERC) consolidator grant for this work.

For more information about this seminar and its speaker, you can visit https://ee380.stanford.edu/Abstracts/...

Support for the Stanford Colloquium on Computer Systems Seminar Series provided by the Stanford Computer Forum.

Colloquium on Computer Systems Seminar Series (EE380) presents the current research in design, implementation, analysis, and use of computer systems. Topics range from integrated circuits to operating systems and programming languages.

It is free and open to the public, with new lectures each week.

Learn more: http://bit.ly/WinYX5

0:00 Introduction
0:51 IBM Research - Zurich
3:03 The Al revolution
3:28 The computing challenge
5:32 Advances in von Neumann computing Storage class memory
7:34 Beyond von Neumann: In-memory computing
9:57 Constituent elements of computational memory
13:17 Multi-level storage capability
14:01 Rich dynamic behavior
15:23 Logic design using resistive memory devices
16:55 Stateful logic
19:04 Bulk bitwise operations
23:49 Matrix-vector multiplication
25:33 Storing a matrix element in a PCM device
26:56 Scalar multiplication using PCM devices
28:32 Application: Compressed sensing and recovery
29:42 Compressed sensing using computational memory
31:39 Compressive imaging: Experimental results
37:20 Crystallization dynamics in PCM
38:38 Example 1: Finding the factors of numbers
40:56 Finding the factors of numbers in parallel
41:51 Example 2: Unsupervised learning of correlations
42:23 Realization using computational memory
44:22 Experimental results (1 Million PCM devices) Device conductance
45:45 Comparative study
51:26 The challenge of imprecision!
54:23 Application 1: Mixed-precision linear solver
55:54 Mixed-precision linear solver: Experimental results
57:28 Application to gene interaction networks
59:28 Application 2: Training deep neural networks