No abstract

In chemoinformatics, drug discovery helps the chemists find new drugs with minimum side effects. The new drug should have similar behavior in terms of the chemical molecular formula to a known drug by applying similarity searching algorithms. A lot of previous works speedup the mentioned task by applying a predefined function, popc, which is only available on new GPUs. This research attempt to provide optimized design using different parallel hardware such as Multicore and GPU processors. Here the aim is to minimize hardware memory allocation by proposed data conversion method and improve data transmission speed. We achieved significant results in terms of performance and execution time in both CUDA and OPENMP designs of fast population count method with data conversion when compared to the sequential code.