A Machine Learning Framework with Feature Selection for Floorplan Acceleration in IC Physical Design

作     者：Shu-Zheng Zhang Zhen-Yu Zhao Chao-Chao Feng Lei Wang Shu-Zheng Zhang;Zhen-Yu Zhao;Chao-Chao Feng;Lei Wang

作者机构：College of Computer Science and TechnologyNational University of Defense TechnologyChangsha 410003China 

基　　金：This work was supported by the HeGaoJi Program of China under Grant Nos.2018ZX01029103 and 2017ZX01038104-002 the National Natural Science Foundation of China under Grant Nos.61802427 and 61902408 

出 版 物：《Journal of Computer Science & Technology》 (计算机科学技术学报（英文版）)

年 卷 期：2020年第35卷第2期

页      码：468-474页

摘      要：Floorplan is an important process whose quality determines the timing closure in integrated circuit(IC)physical *** generating a floorplan with satisfying timing result is time-consuming because much time is spent on the generation-evaluation *** machine learning to the floorplan stage is a potential method to accelerate the floorplan ***,there exist two challenges which are selecting proper features and achieving a satisfying model *** this paper,we propose a machine learning framework for floorplan acceleration with feature selection and model stacking to cope with the challenges,targeting to reduce time and effort in integrated circuit physical ***,the proposed framework supports predicting post-route slack of static random-access memory(SRAM)in the early floorplan ***,we introduce a feature selection method to rank and select important *** both feature importance and model accuracy,we reduce the number of features from 27 to 15(44%reduction),which can simplify the dataset and help educate novice ***,we build a stacking model by combining different kinds of models to improve *** 28 nm technology,we achieve the mean absolute error of slacks less than 23.03 ps and effectively accelerate the floorplan process by reducing evaluation time from 8 hours to less than 60 *** on our proposed framework,we can do design space exploration for thousands of locations of SRAM instances in few seconds,much more quickly than the traditional *** practical application,we improve the slacks of SRAMs more than 75.5 ps(177%improvement)on average than the initial design.

主 题 词：physical design machine learning feature selection design space exploration 

学科分类：081203[081203] 08[工学] 0835[0835] 0812[工学-测绘类] 

核心收录：

D　O　I：10.1007/s11390-020-9688-x

馆 藏 号：203931612...