The recent development of information and communication technology demands high-performance computing for processing a large amount of data. Since data transfer between CPU and storage becomes a bottleneck in some applications, only reducing the operation time on CPU cannot achieve the improvement of computing performance. One of the solutions to this problem is introducing “in-storage computing,” which offloads some operations on an accelerator equipped in storage. By utilizing this technique, we can reduce data transfer between CPU and storage, and hence improve entire system performance. In-storage computing has been applied to machine learning and database systems in prior works. Our laboratory aims to realize an innovative in-storage computing architecture that speeds up the processing of large-scale data.