FPGAs as Computing Platform
Thomas Zink · KOPS (University of Konstanz) · 2008
Since the discovery of Moore's Law advances in hardware have shown to fulfill it accurately. However, maintaining that fact seems to be increasingly difficult. The von Neumann architecture suffers from drawbacks that have become known as the von Neumann Syndrome. Inefficient data transmission through the memory wall and instruction-stream-based sequencing are the source of its illness. To reduce the symptoms the number of computing cores and the cache sizes are increased. But higher parallelism demands greater programming skills. Reconfigurable Computing promises to overcome those problems by performing a paradigm shift towards data-stream-based, instruction-less machines. FPGAs present a low-cost reconfigurable computing platform. This abstract gives a short overview of how FPGAs are used as computing platforms today.