QAMSS Lecture: Precision Materials Engineering for Future Compute and Memory

I will introduce the current compute memory hierarchy and our research on new memory materials and devices for AI, focusing on resistive switching. As AI demands higher compute performance, existing CMOS memory faces a trade-off between speed, endurance, volatility and programming energy. Resistive switching can address these limits by improving memory performance and enabling in-memory computing for neuromorphic AI. By engineering materials from the atomistic level upwards, we have developed quasi-analog, uniform, high-endurance memory cells with tuneable volatility and non-volatility.

About the speaker

Dr. Markus Hellenbrand, University of Cambridge, UK.