FPGA acceleration of large-scale database searches

点击翻译按钮获取中文摘要

The world of computing is changing as the amount of data being processed is growing rapidly and the energy required to do the processing becomes a limitation. At the same time, there is a constant desire for increased performance. Technological limitations make it difficult to continue only with the tried and true processing model using standard microprocessors. To make the required improvements, the computing world has started to use more specialized processors. Graphics processing units (GPUs) are the most common computing unit used after microprocessors. GPUs are particularly effective in applications that have very structured computations. More recently, it has been shown that field-programmable gate arrays (FPGAs), a programmable hardware chip technology, can also be used effectively for some types of computations. This project explores the use of FPGAs in one specific application area. In today's world of big data, there is a need to search extremely large databases. These searches are run using large data centres, where power and performance are especially important in terms of operation costs and the responsiveness to the customer. FPGAs have been shown to be effective in small-scale database systems. This project will integrate FPGAs into a large-scale database application and quantify the benefits of FPGAs for very large database searches.