Abstract—In recent years, graphics processing units (GPU) have become a standard part of high-performance computing systems used for solving large scale computation problems. To relieve the main processor more and more time consumptive tasks are moved from CPU to GPU where algorithms run in parallel on a high number of GPU's processors. In this paper we present both sequential and parallel implementations of a simple matrix multiplication algorithm and we compare the overall execution time. To further speed up the execution we introduce the GPU's fast shared memory and the implementation of the matrix multiplication algorithm that exploits this memory. The results presented in this paper show that the GPU implementation with the use of shared memory is two times faster than the implementation that uses only device's global memory and up to 7.5 times faster than the CPU implementation.

Index Terms—GPU v.s. CPU, CUDA, shared memory, algorithm optimization.

The authors are with University of Ljubljana, Faculty of computer and information science, Slovenia.

Cite:Tomaž Dobravec, Patricio Bulić, "Comparing CPU and GPU Implementations of a Simple Matrix Multiplication Algorithm," International Journal of Computer and Electrical Engineering vol. 9, no. 2, pp. 430-438, 2017.