THERMOS: Thermally-Aware Multi-Objective Scheduling of AI Workloads on Heterogeneous Multi-Chiplet PIM Architectures
Open Access
- 26 September 2025
- journal article
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Embedded Computing Systems
- Vol. 24 (5s) , 1-26
Abstract
Chiplet-based integration enables large-scale systems that combine diverse technologies, enabling higher yield, lower costs, and scalability, making them well-suited to AI workloads. Processing-in-Memory (PIM) has emerged as a promising solution for AI inference, leveraging technologies such as ReRAM, SRAM, and FeFET, each offering unique advantages and tradeoffs. A heterogeneous chiplet-based PIM architecture can harness the complementary strengths of these technologies to enable higher performance and energy efficiency. However, scheduling AI workloads across such a heterogeneous system is challenging due to competing performance objectives, dynamic workload characteristics, and power and thermal constraints. To address this need, we propose THERMOS, a thermally-aware, multi-objective scheduling framework for AI workloads on heterogeneous multi-chiplet PIM architectures. THERMOS trains a single multi-objective reinforcement learning (MORL) policy that is capable of achieving Pareto-optimal execution time, energy, or a balanced objective at runtime, depending on the target preferences. Comprehensive evaluations show that THERMOS achieves up to 89% faster average execution time and 57% lower average energy consumption than baseline AI workload scheduling algorithms with only 0.14% runtime and 0.022% energy overhead.Keywords
All Related Versions
This publication has 35 references indexed in Scilit:
- Neural architecture search for in-memory computing-based deep learning acceleratorsNature Reviews Electrical Engineering, 2024
- First demonstration of in-memory computing crossbar using multi-level Cell FeFETNature Communications, 2023
- A practical guide to multi-objective reinforcement learning and planningAutonomous Agents and Multi-Agent Systems, 2022
- Colonnade: A Reconfigurable SRAM-Based Digital Bit-Serial Compute-In-Memory Macro for Processing Neural NetworksIEEE Journal of Solid-State Circuits, 2021
- A 7-nm Compute-in-Memory SRAM Macro Supporting Multi-Bit Input, Weight and Output and Achieving 351 TOPS/W and 372.4 GOPSIEEE Journal of Solid-State Circuits, 2020
- A Programmable Heterogeneous Microprocessor Based on Bit-Scalable In-Memory ComputingIEEE Journal of Solid-State Circuits, 2020
- UNPU: An Energy-Efficient Deep Neural Network Accelerator With Fully Variable Weight Bit PrecisionIEEE Journal of Solid-State Circuits, 2018
- Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference2018
- Scalarized multi-objective reinforcement learning: Novel design techniques2013
- Multicore thermal management with model predictive control2009