{"title":"An effective method to use GPU for rectangle packing","authors":"M. Kataoka, Shuji Tsukiyamia, T. Kambe, M. Fukui","doi":"10.1109/NEWCAS.2012.6328973","DOIUrl":null,"url":null,"abstract":"As useful software tools for GPU (Graphics Processing Unit) have been provided, GPGPU (General Purpose computing on GPU) becomes popular recently. However, the usage of GPU in the field of CAD algorithm for VLSI has not been large yet. The reason of this is that algorithms in this field have a tendency such that they must change operations of the next step according to the result of the current step. Hence, introduction of parallelism is not easy. In this paper, we consider rectangle packing problem which is important to VLSI floorplan, and propose an algorithm which can employ GPU effectively. The algorithm is obtained by improving defects of temperature parallel simulated annealing algorithm. Experimental results for benchmark data show its effectiveness.","PeriodicalId":122918,"journal":{"name":"10th IEEE International NEWCAS Conference","volume":"27 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2012-06-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"10th IEEE International NEWCAS Conference","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/NEWCAS.2012.6328973","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
As useful software tools for GPU (Graphics Processing Unit) have been provided, GPGPU (General Purpose computing on GPU) becomes popular recently. However, the usage of GPU in the field of CAD algorithm for VLSI has not been large yet. The reason of this is that algorithms in this field have a tendency such that they must change operations of the next step according to the result of the current step. Hence, introduction of parallelism is not easy. In this paper, we consider rectangle packing problem which is important to VLSI floorplan, and propose an algorithm which can employ GPU effectively. The algorithm is obtained by improving defects of temperature parallel simulated annealing algorithm. Experimental results for benchmark data show its effectiveness.
随着图形处理单元(Graphics Processing Unit, GPU)软件工具的不断涌现,GPGPU (General Purpose computing on GPU)成为近年来的热门技术。然而,GPU在超大规模集成电路CAD算法领域的应用还不是很大。这是因为该领域的算法有一种倾向,即必须根据当前步骤的结果改变下一步的操作。因此,引入并行性并不容易。本文针对超大规模集成电路平面设计中的矩形布局问题,提出了一种有效利用GPU的算法。该算法是通过改进温度并行模拟退火算法的缺陷而得到的。对基准数据的实验结果表明了该方法的有效性。