X-OPM improves on-chip power modeling accuracy and robustness
X-OPM: Explainable Automatic Digital On-Chip Power Modeling for Enhanced Robustness
Hardware ArchitectureArtificial IntelligenceMachine Learning
Summary
Predicting the power used by a computer chip while it is running helps save energy and improve performance. Existing methods for measuring this power often work well only on known types of tasks and can be hard to explain. The authors introduce X-OPM, which uses knowledge about how chips work to build simpler and more reliable models that can explain how different factors affect power use. This new method works better across different tasks and uses very little extra chip space.
What this means in practice
- •For processor designers: Integrate an explainable and lightweight power meter into chips to improve real-time power prediction and management accuracy.
- •For embedded systems developers: Use X-OPM to enable more robust and efficient power management on commercial vector processors with minimal hardware overhead.
- •For data center operators: Deploy hardware using X-OPM to improve power-aware scheduling, enhancing energy efficiency across diverse workloads.$Commercial implications: Allows the sale of energy-efficient servers with built-in reliable power meters for improved workload management.
Authors
Jingbo Jiang, Xizi Chen, Jian Peng, Wei Zhang
Abstract
Proactive power management systems reduce processor dynamic power through runtime power prediction and power-aware scheduling. Accurate, stable and low-overhead digital on-chip power meters (OPMs) are crucial for improving the prediction quality. Recent studies have explored various modeling methods, including using linear models, decision trees, and multi-layer perceptrons (MLPs) to construct OPMs. However, most current approaches train models end-to-end without analyzing the physical interpretability of features, affecting their ability to generalize to unseen workloads. Grounded in the design principles of synchronous digital VLSI circuits, X-OPM introduces a robust feature engineering framework that uses tree-based models to capture feature interactions and linear models for prediction. It also incorporates a human-in-the-loop workflow to balance model accuracy against modeling effort. Evaluated on a commercial C906 vector processor, X-OPM consistently achieves $R^2 > 0.93$ across all workloads with sampling window size set below $8$ cycles. In contrast, state-of-the-art methods including APOLLO, COBIT, and standard MLPs fail to generalize across all test cases. Layout with commercial EDA tools shows that X-OPM incurs an area overhead below $0.1\%$, which is on par with lightweight tree-based and linear models, and significantly smaller than MLP-based models.