With the advancement of graphics hardware, graphics processor units (or GPUs) are becoming better suited for general purpose computing. These massively parallel computing devices are used in a wide array of applications, for example in bioinformatics (DNA-sequencing, protein folding), medicine (molecular dynamics) and finances (risk analysis). GPUs are becoming a prevalent solution for tackling time intensive computing problems, and a transition to high degree parallelism can be observed in the CPU (Central Processing Unit) industry, too.
With the advancement of information technology and the rapid expansion of computer networks, protection against malware (malicious software) has become an increasingly important task. With the fast increase of network bandwidth and data storage space, scanner applications capable of high data throughput are becoming extremely important.
The question emerges: is the GPU suited to provide effective protection against malicious software? GPUs, while providing tremendous computing capabilities, also introduce a new architecture, thus requiring new angles of approach for the efficient utilization of their capabilities. To answer this question, the specialized implementation of an effective GPU-specific algorithm is required.
This paper presents a signature-based detection algorithm, which – thanks to its construction – is capable of harnessing the massive parallelism of the GPU, thus hardware accelerating malware detection.