Experimental evaluations of high-level energy optimization based on thread partitioning

This work presents a thread partitioning algorithm for high-level synthesis systems which generate low energy circuits. In the algorithm, we partition a thread into two sub-threads, one of which has RF and the other does not have RF. The partitioned sub-threads need to be synchronized with each other to keep the data dependency of the original thread. Since the partitioned sub-threads have waiting time for synchronization, gated clocks can be applied to each sub-thread. We achieve 33% energy reduction when we apply our proposed algorithm to a JPEG encoder.