GPU-Driven Scalable Parser for OBJ Models

作     者：Sunghun Jo Yuna Jeong Sungkil Lee 

作者机构：Department of Software Sungkyunkwan University Suwon 16419 Korea 

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

年 卷 期：2018年第33卷第2期

页      面：417-428页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 08[工学] 081201[工学-计算机系统结构] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：supported in part by the Mid-career and Global Frontier (on Human-centered Interaction for Coexistence) Research Development Programs through the National Research Foundation (NRF) under Grant the Information Technology Research Center (ITRC) Program under Grant supervised by the Institute for Information and Communications Technology Promotion (IITP) funded by the Korea Government (Ministry of Science, ICT (Information and Communications Technologies) and Future Planning) Faculty Research Fund, Sungkyunkwan University, 2011 

主　　题：3D model Wavefront OBJ parser GPU 

摘      要：This paper presents a scalable parser framework using graphics processing units （GPUs） for massive text-based files. Specifically, our solution is designed to efficiently parse Wavefront OBJ models texts of which specify 3D geometries and their topology. Our work bases its scalability and efficiency on chunk-based processing. The entire parsing problem is subdivided into subproblems the chunk of which can be processed independently and merged seamlessly. The within-chunk processing is made highly parallel, leveraged by GPUs. Our approach thereby overcomes the bottlenecks of the existing OBJ parsers. Experiments performed to assess the performance of our system showed that our solutions significantly outperform the existing CPU-based solutions and GPU-based solutions as well.