- Abstract The increase in performance of the last generations of graphics processors (GPUs) has made this class of platform a co-processing tool with remarkable success in certain types of operations. - In this thesis, I focused to put out the solutions to build up a desktop with many GPUs to implement parallel processing applications with a lot of input data, then testing and evaluating the performance of the Level 3 operations of BLAS and LinPack Benchmark using CUBLAS library based on NVIDIA’s GPUs with unified architecture support. - From this study, I gain insights on the quality of the kernels in the library and I propose several alternative implementations that are competitive with those in CUBLAS. - Experimental results on a GeForce GTX 295 compare the performance of GPU and CPU.
Xem thử không khả dụng, vui lòng xem tại trang nguồn hoặc xem
Tóm tắt