{"title":"Scalable rendering of large SPH simulations using an RK-enhanced interpolation scheme on constrained datasets","authors":"K. Griffin, C. Raskin","doi":"10.1109/LDAV.2016.7874335","DOIUrl":null,"url":null,"abstract":"Smoothed particle hydrodynamics (SPH) is a Lagrangian alternative to mesh-based schemes for simulating fluid flows in a wide variety of physical applications. However, there are a number of challenges that arise when attempting to visualize the results of these simulations. This poster presents a Reproducing Kernel (RK) enhanced SPH Resample Operator we have developed, in VisIt [1], to run on high performance computing (HPC) platforms, scale across compute nodes, and work efficiently on constrained datasets. We define constrained datasets as data that is exported in a way as to not allow efficient processing within a visualization tool. For us, these constrained datasets are pre-partitioned spatially, which in most cases, is not ideal for good load balancing. Furthermore, they also lack metadata, like the identification of halo or ghost zone regions, needed for node independent processing.","PeriodicalId":148570,"journal":{"name":"2016 IEEE 6th Symposium on Large Data Analysis and Visualization (LDAV)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2016-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2016 IEEE 6th Symposium on Large Data Analysis and Visualization (LDAV)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/LDAV.2016.7874335","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
Smoothed particle hydrodynamics (SPH) is a Lagrangian alternative to mesh-based schemes for simulating fluid flows in a wide variety of physical applications. However, there are a number of challenges that arise when attempting to visualize the results of these simulations. This poster presents a Reproducing Kernel (RK) enhanced SPH Resample Operator we have developed, in VisIt [1], to run on high performance computing (HPC) platforms, scale across compute nodes, and work efficiently on constrained datasets. We define constrained datasets as data that is exported in a way as to not allow efficient processing within a visualization tool. For us, these constrained datasets are pre-partitioned spatially, which in most cases, is not ideal for good load balancing. Furthermore, they also lack metadata, like the identification of halo or ghost zone regions, needed for node independent processing.