Gordon Reid

GPU-Accelerated Document Filtering in OpenCL


Document filtering a is problem of growing importance. The advent of the Internet has resulted in a massive surge in the transfer and storage of data and information. A prime example is email where hundreds of billions are sent every day. Emails, and other kinds of documents, require to be filtered. This problem is highly data parallel in nature as the classification of one document does not affect the classification of another. The project investigates the viability of using GPUs rather than CPUs for the classification. GPUs are designed with parallel problems in mind and thus have the potential to allow for significantly higher throughput compared with CPUs.