Applying Binary Code Similarity Detection on Acceleration Processor

Binary code similarity detection, as an integral part of network security, has remained a hot spot in the research fields of intellectual property protection, vulnerability detection, patch analysis, and malicious software detection. It carries significant usability value. With the rapid development of deep learning and neural network technologies, new algorithms and thoughts have been introduced into the research of binary code similarity detection, achieving breakthrough progress. MT-3000, a heterogeneous multi-zone acceleration processor, which is entirely designed and implemented by National University of Defense Technology, possesses powerful computing performance. In this paper, based on the latest deep learning theory and using the domestically produced and controllable high-performance MT-3000 acceleration processor for relevant tests and researches on the binary code similarity detection, proposes relevant conclusions and suggestions.