GPGPU AND SIEVE OF SUNDARAM

作     者：Ghidarcea, Mircea Popescu, Decebal 

作者机构：UNST POLITEHNICA of Bucharest – Computer Science Romania 

出 版 物：《UPB Scientific Bulletin, Series C: Electrical Engineering and Computer Science》 (UPB Sci. Bull. Ser. C Electr. Eng. Comput. Sci.)

年 卷 期：2024年第2024卷第4期

页      面：31-42页

核心收录：

主　　题：Optimization algorithms 

摘      要：In the quasi-forgotten universe of prime sieving, Sieve of Sundaram is seen as an elegant approach to prime generation, but its performance is significantly worse than basically every other algorithm but brute-force trial-division. Yet, exactly this less than stellar arithmetic complexity makes it an ideal candidate to demonstrate the power of GPU based computing in tackling this kind of problems that evade the usual matrix oriented, symmetrical jobs for which the GPGPU paradigm was natively devised. This paper intends to demonstrate how a well thought approach can benefit from the immense throughput of modern GPUs to solve problems like sieving, which are not perfectly symmetric. © 2024, Politechnica University of Bucharest. All rights reserved.