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

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhang, Qijun, Li, Mengming, lu, Yao, Xie, Zhiyao
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917813159985152
author Zhang, Qijun
Li, Mengming
lu, Yao
Xie, Zhiyao
author_facet Zhang, Qijun
Li, Mengming
lu, Yao
Xie, Zhiyao
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2410_17789
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle FirePower: Towards a Foundation with Generalizable Knowledge for Architecture-Level Power Modeling
Zhang, Qijun
Li, Mengming
lu, Yao
Xie, Zhiyao
Hardware Architecture
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.
title FirePower: Towards a Foundation with Generalizable Knowledge for Architecture-Level Power Modeling
topic Hardware Architecture
url https://arxiv.org/abs/2410.17789