Abstract not found

Data referencing during program execution can be a signicant source of energy consumption especially for data-intensive programs. In this paper, we propose an approach to minimize such energy consumption by allocating data to proper registers and memory. Through careful analysis of boundary conditions between consecutive blocks, our approach eÆciently handles various control structures including branches, merges and loops, and achieves the allocation results beneting the whole program. The computational cost for solving the allocation problem is rather low comparing with known approaches while the quality of the results are very encouraging. 1 Introduction Today's high demand of portable electronic products makes low energy consumption as important as high speed and small area in computer system design. Even for non-portable high performance systems, lower power consumption design helps to decrease packing and cooling cost and increase the reliability of the systems [25]. A lot of r...

Data referencing during program execution can be a significant source of energy consumption especially for data-intensive programs. In this paper, we propose an approach to minimize such energy consumption by allocating data to proper registers and memory. Through careful analysis of boundary conditions between consecutive blocks, our approach efficiently handles various control structures including branches, merges and loops, and achieves superior allocation results for the whole program. The computational cost of our approach for solving the global register allocation problem is rather low comparing with known approaches while the quality of our results is very encouraging.