Comparision of graphics processing units and central processing units

Graphic processors are becoming faster and faster. Computational power within graphic processing units (GPUs) is growing rapidly compared to central processing units (CPUs). Usage of this power is becoming very interesting in many areas. Programmers try to use this power. They are developing new algorithms for non-graphic applications. When we do not play games, GPUs are idle and this is most of the time. Parallel processing algorithms which exploit both GPUs and CPUs takes place here. More I was researching this area, more interesting it was getting. General purpose computing on graphic processors (GPGPU) is relatively new and it is available to everyone. This is the main reason why it is developing so fast. In this thesis I tried to represent background and developing of both graphic and main processors through time. I presented architecture on general examples and on specific processors which are widely used in personal computers. I brought this theme to the close with analyzing execution of matrix multiplication program. I measured time needed for execution of the program on CPU and on the GPU. As example I used Intel's Core 2 Duo processor E7400 and NVIDIA's graphic card GTX260. Speedups in applications were up to 300 times on GPU. I worked with NVIDIA’s environment CUDA, based on C programming language. With CUDA, it is possible to unlock the processing power of the GPU to solve complex compute-intensive problems. Environment is easy to integrate with Microsoft Visual Studio and is easy to use