high performance computing on graphics processing units: hgpu.org

Graphics processors (GPU – Graphic Processor Units) recently have gained a lot of interest as an efficient platform for general-purpose computation. Cellular Automata approach which is inherently parallel gives the opportunity to implement high performance simulations. This paper presents how shared memory in GPU can be used to improve performance for Cellular Automata models. In our previous works, we proposed algorithms for Cellular Automata model that use only a GPU global memory. Using a profiling tool, we found bottlenecks in our approach. With this paper, we will introduce modifications that takes an advantage of fast shared memory. The modified algorithm is presented in details, and the results of profiling and performance test are demonstrated. Our unique achievement is comparing the efficiency of the same algorithm working with a global and shared memory.