摘要
针对"排序网络"的特点及其广泛运用的意义,依据现代GPU在通用计算方面的功能,提出了基于Brook在GPU上执行通用计算的实现方法,并对"排序网络"在GPU的实现进行了测试,结果表明GPU在Brook环境下通用计算能力优于CPU,且排序网络的并行计算的特点能够在GPU上充分展现.
The general computing method based on Brook is proposed in the paper according to general computing function of modern GPU, characteristic and extensive use of the ＂Sorting Network＂.The testing result of ＂sorting network＂ by GPU＇s realization presents the computing ability of GPU has an advantage over CPU and parallel computing characteristic can carried out on GPU under the Brook environment.
出处
《微电子学与计算机》
CSCD
北大核心
2010年第5期53-56,共4页
Microelectronics & Computer
基金
国家自然科学基金项目(60873030)
关键词
排序网络
数据并行计算
流计算
GPU通用计算
Brook
sorting network
data parallel computing
stream computing
GPU general purpose computation
brook
