GPU involvement in non-graphical calculations and data processing

This research study is based on the growing interest towards graphical processing unit usability for non-graphical calculations and data processing tasks. The aim of the study is to implement basic mathematical and data processing algorithms on graphical processor with CUDA architecture. Through employment of various programs, written in CUDA C for GPU and C programming language for CPU, similar mathematical and data managing tasks were tested on both processing units and comparison of processors performance was presented. Also algorithms and applications, developed specifically for implementation on parallelized architecture of CUDA enabled GPUs, were shown in this research. The importance of parallelism invocation for task processing on graphical hardware was recognized and optimization techniques for parallelism usage were studied. This research reveals how the usage of GPU speeds up a process of mathematical calculations and influences on data processing tasks performance. It as well illustrates features and limitations in graphical processor use, and determines tasks and data quality/quantity levels, where GPU usage is the most significant