Prenaut: Design space exploration for embedded symmetric multiprocessing with various on-chip architectures

详细信息    查看全文

• 作者：Gorker Alp Malazgirt ; ^{alp.malazgirt@boun.edu.tr}Author Vitae ; Arda Yurdakul ^{yurdakul@boun.edu.tr}Author Vitae
• 关键词：Design space exploration ; Symmetric multiprocessing ; Machine learning ; Clustering
• 刊名：Journal of Systems Architecture
• 出版年：2017
• 出版时间：January 2017
• 年：2017
• 卷：72
• 期：Complete
• 页码：3-18
• 全文大小：2320 K
• 卷排序：72

文摘

Prenaut is a design space exploration method for finding the best on-chip architectures given processor cores, Level 1, Level 2, and Level 3 caches. Current design space exploration methods mostly explore cache, processor configurations with a fixed architecture. Our method explores architectures with fixed cache and processor configurations. Prenaut is to build a data oriented design space exploration method that exploits simulation data to its full extent rather than discarding it. Prenaut uses the available simulation data and applies machine learning methods for estimating design parameters.