International Journal of Information Studies Mustafa AL-Barmawi, Nurul Malim 5 3 2000 https://www.dline.info/ijis/fulltext/v5n3/ijisv5n3_3.pdf Molecular similarity searching is a process to find chemical compounds that are similar to a target compound. The concept of molecular similarity plays an important role in modern computer aided drug design methods, and has been successfully applied in the optimization of lead series. Similarity Fusion (SF) is one of the known methods in Virtual Screening. It uses 2D fingerprint for measuring the structural similarity. Although SF introduces a simple way in estimating the degree of structural similarity between two molecules, it is actually effective and can be considered an efficient tool for searching large chemical databases with the existence of multiple similarity coefficients. The diversity in measures has shown that employment of data fusion rule in fingerprint-based similarity searching can improve results over the traditional search. In this work, the results are referring to the similarity scores between the reference structure and the database compound. The problem while performing Similarity Fusion is that it consumes a lot of time as a result of the increasing number of coefficients. It becomes even more time consuming when it fuses the multi reference similarity searching. This study aims to parallelize the SF for an improved execution speed and show the impact of parallelism on the execution time. In this work, to overcome the dependency between cells, Tiling Algorithm was firstly applied followed by parallelization of SF on the Multicore and General Purposes Graphical Processing Unit (GPGPU). However, the results obtained proved otherwise especially when dealing with GPU platform. The results are submitted that Multicore processing provides better results.