Computing In Spintronic Memory: A Thermal Perspective

2026-04-08

AI Summary

This work presents the first systematic quantitative analysis of thermal characteristics in spintronic compute-in-memory (CiM) architectures, addressing the thermal hotspots induced by high activity levels that exacerbate power density despite CiM’s potential to mitigate the memory wall. By integrating thermal conduction modeling, spintronic non-volatile memory array simulation, and microarchitectural power analysis, the study reveals that temperature distribution across CiM arrays remains largely uniform. It further demonstrates that peak temperature scales linearly with the number of active computing units and inversely with array size, while the underlying memory technology critically governs both power density and thermal behavior. These findings establish key quantitative relationships among workload intensity, physical dimensions, device technology, and thermal profiles in spintronic CiM systems.

Abstract

Computing-in-Memory (CiM) is a promising paradigm to address the memory bottleneck constraining traditional systems. Most power-efficient CiM variants can directly perform Boolean operations in non-volatile memory arrays. Higher microarchitectural activity due to CiM, however, can significantly increase power density (power per area) and result in thermal hotspots. In this paper, we provide a quantitative thermal characterization for CiM. We demonstrate that (i) the temperature remains mostly uniform due to lateral thermal conduction; (ii) the temperature increases linearly with the number of memory cells participating in computation; (iii) the temperature decreases linearly with the memory array size; (iv) the memory technology dictates the power density, hence the thermal characteristics.

Problem

• Computing-in-Memory
• thermal hotspots
• power density
• spintronic memory
• thermal characterization

Innovation

• Computing-in-Memory
• thermal characterization
• spintronic memory
• power density
• thermal hotspot