关键词：自然语言应用程序；图像处理器；绘图处理器（GPUs）；自然语言解析
摘 要：As we enter the era of mobile computing, high-quality and efficientnatural language applications become more and more important,which greatly facilitate intelligent human-computer interaction.Unfortunately, most high-quality natural language applicationsemploy large statistical models, which render them impractical forreal-time use. Meanwhile, Graphics Processor Units (GPUs) have become widely available, offering the opportunity to alleviate thisbottleneck by exploiting the fine-grained data parallelism found inthe natural language processing algorithms. In this report, weexamine the possibility of parallelizing two major natural languageapplications, natural language parsing and machine translation onGPUs. In natural language parsing, we explore the design space ofparallelizing the dynamic programming computations carried outby the CKY parsing algorithm. We use the Compute Unified Device Architecture (CUDA) programming model to re-implementa state-of-the-art parser, and compare its performance on tworecent GPUs with different architectural features.