Kaushik Roy at Purdue
video 2020 Talk 1: In-Memory Computing based Machine Learning Accelerators: Opportunities and Challenges Dr. Kaushik Roy
https://engineering.purdue.edu/NRL/Research
2020 Pathways to efficient neuromorphic computing with non-volatile memory technologies
In this paper, we focus on non-volatile memory technologies and their applications to bio-inspired
neuromorphic computing, enabling spike-based machine intelligence
....
Architecturally, such crossbars can be connected in a distributed manner, bringing
in additional system-level parallelism, a radical departure from the conventional von-Neumann architecture.
As discussed earlier, spin devices suffer from very low ON/OFF resistance ratios compared to other technologies. Hence, despite experimental demonstration of isolated synaptic spin devices,119 large-scale crossbar-level neuromorphic implementations have been mostly limited to simulation studies
open source - PUMA: A Programmable Ultra-efficient Memristor-based Accelerator for Machine Learning Inference ankit 2019
https://github.com/illinois-impact/puma-compiler
https://github.com/Aayush-Ankit/puma-simulator
-
chakraborty 2018 Technology Aware Training in Memristive Neuromorphic Systems for Nonideal Synaptic Crossbars
PUMA uses ROM-embedded SRAM - 2012 Area Efficient ROM-Embedded SRAM Cache
2018 SPARE: Spiking Neural Network Acceleration Using ROM-Embedded RAMs as In-Memory-Computation Primitives Amogh Agrawal
2014 SPINDLE: SPINtronic Deep Learning Engine for Large-scale Neuromorphic Computing
Computing in Memory With Spin-Transfer TorqueMagnetic RAM
-
X-SRAM: Enabling In-Memory Boolean Computations in CMOS Static Random Access Memories
-
DSTT-MRAM: Differential Spin Hall MRAM for On-chip Memories Yusung Kim
