FirePower: Towards a Foundation with Generalizable Knowledge for Architecture-Level Power Modeling

TL;DR

FirePower introduces a novel architecture-level power modeling approach that leverages cross-architecture knowledge and component-level models to achieve high accuracy with minimal training data, aiding early CPU design optimizations.

Contribution

It presents a new power modeling framework that enables effective few-shot learning for new architectures by separating generalizable and architecture-specific knowledge.

Findings

Achieves 5.8% error with only two configurations.

Outperforms baseline by 8.8% in error reduction.

High correlation coefficient of 0.98 indicating accurate modeling.

Abstract

Power efficiency is a critical design objective in modern processor design. A high-fidelity architecture-level power modeling method is greatly needed by CPU architects for guiding early optimizations. However, traditional architecture-level power models can not meet the accuracy requirement, largely due to the discrepancy between the power model and actual design implementation. While some machine learning (ML)-based architecture-level power modeling methods have been proposed in recent years, the data-hungry ML model training process requires sufficient similar known designs, which are unrealistic in many development scenarios. This work proposes a new power modeling solution FirePower that targets few-shot learning scenario for new target architectures. FirePower proposes multiple new policies to utilize cross-architecture knowledge. First, it develops power models at component level, and components are defined in a power-friendly manner. Second, it supports different generalization strategies for models of different components. Third, it formulates generalizable and architecture-specific design knowledge into two separate models. FirePower also supports the evaluation of the generalization quality. In our experiments, FirePower can achieve a low error percentage of 5.8% and a high correlation R of 0.98 on average only using two configurations of target architecture. This is 8.8% lower in error percentage and 0.03 higher in R compared with directly training McPAT-Calib baseline on configurations of target architecture.