Speeding up Online Placement for XILINX FPGAs by Reducing Configuration Overhead.
Ali Ahmadinia, Jürgen Teich · 2003
Recent generations of FPGAs allow run-time partial reconfiguration. To increase the efficacy of reconfigurable computing, multitasking on FPGAs is proposed. One of the challenging problems in multitasking systems is online template placement. In this paper, we describe how existing algorithms work, and propose a new multi-stage method for mapping of tasks to reconfigurable hardware. Also a new fitting algorithm, as a part of online placement, has been developed. This method tries to reduce the reconfiguration overhead. First simulation results are promising.