Using Application Bisection Bandwidth to Guide Tile Size Selection for the Synchroscalar Tile-Based Architecture

This paper investigates the impact of proper tile size selection on the power the power consumption for tile-based processors. We refer to this investigation as a tile granularity study. This is accomplished by distilling the architectural cost of tiles with different computational widths into a system metric we call the Granularity Indicator (GI). The GI is then compared against the bisection bandwith of algorithms when partitioned across multiple tiles. From this comparison, the tile granularity that best fits a given set of algorithms can be determined, reducing the system power for that set of algorithms. When the GI analysis is applied to the Synchroscalar tile architecture[1], we find that Synchroscalar’s already low power consumption can be further reduced by 14% when customized for execution of the 802.11a reciever. In addition, the GI can also be a used to evaluate tile size when considering multiple applications simultaneously, providing a convenient platform for hardware-software co-design.