Evaluation of the Parallel Computing Performances based on the nVIDIA GeForce GPU

Parallel programming is a form of computation in which the calculations are carried out simultaneously, operating on the principle where large problems can be divided into smaller, which are then solved in parallel. Most common this programming is used in high performance computing, but due to the physical constraints which prevent frequency scaling the interest is even higher. As computers consumption has become a problem in the recent years, the parallel programming has grown into the dominant paradigm in computer architecture, mainly in the form of multicore processors. The paper shows the process of designing parallel programs and it includes some results obtained by testing the parallel programming performance of different nVIDIA GeForce graphics cards. The purpose of the test is to compare performance of several types of GPUs for various applications. The results of the test could not specifically say which GPU is best due to the different features of the cards, but they can be used to determine which card offers better performances for different parts of the tests