Operation tables for scheduling in the presence of incomplete bypassing

Register bypassing is a powerful and widely used feature in modern processors to eliminate certain data hazards. Although complete bypassing is ideal for performance, bypassing has significant impact on cycle time, area, and power consumption of the processor. Due to the strict constraints on performance, cost and power consumption in embedded processors, architects need to evaluate and implement incomplete register bypassing mechanisms. However traditional data hazard detection and/or avoidance techniques used in retargetable schedulers breaks down in the presence of incomplete bypassing. We present the concept of operation tables, which can be used to detect data hazards, even in the presence of incomplete bypassing. Furthermore our technique integrates the detection of both data, as well as resource hazards, and can be easily employed in a compiler to generate better schedules. Our experimental results on the popular Intel XScale embedded processor platform show that even with a simple intra-basic block scheduling technique, we achieve up to 20% performance improvement over fully optimized GCC generated code on embedded applications from the MiBench suite.