Non-uniform Set-Associative Caches for Power-Aware Embedded Processors

Power consumption is becoming one of the most important constraints for microprocessor design in nanometer-scale technologies. Especially, as the transistor supply voltage and threshold voltage are scaled down, leakage energy consumption is increased even when the transistor is not switching. This paper proposes a simple technique to reduce the static energy due to subthreshold leakage current. The key idea of our approach is to allow the ways within a cache to be accessed at different speeds and to place infrequently accessed data into the slow ways. We use dual-Vt technique to realize the non-uniform set-associative cache, and propose a simple replacement policy to reduce average access latency. Experimental results on 32-way set-associative caches demonstrate that any severe increase in clock cycles to execute application programs is not observed and significant static energy reduction can be achieved, resulting in the improvement of energy-delay2 product.