An adaptive and scalable multiprocessor system For Xilinx FPGAs using minimal sized processor cores
Alexander Klimm, Lars Braun, Jürgen Becker · Proceedings - IEEE International Parallel and Distributed Processing Symposium · 2008
In embedded systems, especially in multi-purpose platforms, the need for computational power can vary greatly. This depends mainly on the type of applications running on the platform as well as on limitations such as real-time constraints or the amount of data being processed. These factors can change during lifetime of the system and even during runtime. A highly adaptive computing platform is desirable, providing exactly the computational power and speed needed for any particular task at any particular time using minimal resources. This contribution discusses a highly scalable multiprocessor system composed of minimal sized processor cores. The proposed architecture is intended for computational intensive applications such as on-line routing and placement on FPGAs or applications in the cryptographic domain. While offering a lot of leeway in computational power it uses minimal resources on an FPGA, thus freeing hardware resources for applications running in parallel to the multiprocessor system.