Software support for dynamic partial reconfigurable FPGAs on heterogeneous platforms

This thesis addresses the design and implementation of a software support for real-time systems developed on heterogeneous platforms that include a processor and an FPGA with dynamic partial reconfiguration capabilities. The software support enables tasks to request the execution of accelerated functions on the FPGA in parallel with other tasks running on the processor. Accelerated functions are dynamically allocated on the FPGA depending of the availability of the area and the online requests issued by the processor, so extending the concept of multitasking to the FPGA resource domain. The performance of the allocation mechanism has been evaluated in terms of speed-up and response times. The achieved results show that the system is able to guarantee bounded delays and acceptable overhead that can be taken into account for a future schedulability analysis of real-time applications