A Comparison of Frameworks for Heterogeneous Computing Using High-Performance Pattern-Matching for DNA Biometrics and Digital Forensics

In this paper we explore popular parallel computing frameworks, used in high-performance pattern-matching applications involving heterogenous architectures, specifically based on CPU/GPU infrastructures, and with particular emphasis on the Aho-Corasick multiple pattern-matching approach and its applications to DNA sequence analysis and digital forensics. Our experiments showed that modern managed frameworks, such as SYCL, could serve as a viable replacement for dedicated frameworks such as CUDA and OpenCL, showcasing a comparable throughput, while being more flexible and developer-friendly.