A Runtime Profiler: Toward Virtualization of Polymorphic Computing Platforms
Hamid Mushtaq, Mojtaba Sabeghi, Koen Bertels · 2010
Runtime multitasking support on Reconfigurable Computers requires complicated resource management techniques in which the FPGA area has to be shared between multiple concurrent tasks dynamically. Such a resource allocation mechanism needs to know the current configuration and load of the system in order to decide about the allocation of the resources. In such systems, a runtime profiler is an important tool which can give vital information about the running applications on the system. In this paper, we present the design and implementation of a runtime profiler which is responsible to produce statistics about the code running on the system. We have performed a set of experiments in order to show the overhead of our proposed profiler. The evaluation results show that the overhead imposed by the profiler is less than 1.5% of the total execution time and the information generated by the profiler is almost as accurate as a design time profiler such as gprof.