Abu Sebastian1 Manuel Le Gallo1 Evangelos Eleftheriou1

1, IBM Research-Zurich, Ruschlikon, , Switzerland

Given the explosive growth in data-centric cognitive computing and the imminent end of CMOS scaling laws, it is becoming increasingly clear that we need to transition to non-von Neumann computing architectures. A first step in this direction could be in-memory computing whereby certain computational tasks are performed in place in a specialized memory unit, which we call computational memory.

Phase-change memory (PCM) devices could play a key role as elements of such a computational memory unit. The physical attributes of these devices can be exploited to achieve in-place computation. When organized in a cross-bar configuration, PCM devices can be used to perform matrix-vector multiplications with very low computational complexity [1]. An appealing application of this concept is for the problem of compressed sensing and recovery of high-dimensional signals [2]. This is an application in which the lack of precision arising from the matrix-vector multiplication operations is not prohibitive. However, in other applications such as solving systems of linear equations or training deep neural networks, the lack of precision could be a key challenge. To address this, we propose the concept of mixed-precision in-memory computing in which, through a judicious combination of high-precision processing units and computational memory, we can achieve arbitrarily high precision, while still retaining much of the benefits of non-von Neumann computing [3]. Finally, I will present applications in which the dynamics of PCM devices, such as the crystallization dynamics and the dynamics of structural relaxation, are used to perform computational tasks. Examples include finding factors of numbers in parallel [4] and the detection of temporal correlations between event-based data streams [5].

[1] G.W. Burr et al., “Neuromorphic computing using non-volatile memory”, Advances in Physics: X, 2:1, 89-124, 2017
[2] M. Le Gallo et al., “Compressed sensing recovery using computational memory”, Proceedings of IEDM, 2017
[3] M. Le Gallo et al., “Mixed precision in-memory computing”, ArXiv, arXiv:1701.04279v3, 2017
[4] P. Hosseini et al., “Accumulation-based computing using phase-change memories with FET access devices”, IEEE Electron Device Letters, 36:9, 975-977, 2015
[5] A. Sebastian et al., “Temporal correlation detection using computational phase-change memory”, Nature Communications, 8, article 1115, 2017