Register Allocation Optimization in a Kaffe Based Dynamic Compilation

Dynamic compilation and optimization are widely used in heterogeneous computing environments, in which an intermediate form of the code is compiled to native code during execution. An important tradeoff exists between the amount of time spent dynamically optimizing the program and the running time of the program. In this paper, we explore this trade-off for an important optimization – global register allocation. I present a graph-coloring register allocation in a Kaffe based JIT that has been redesigned for runtime compilation. Compared to ChaitinBriggs, a standard graph-coloring technique, the reformulated algorithm requires considerably less allocation time and produces allocations that are only marginally worse than those of Chaitin-Briggs. The experimental results indicate that the allocator performs better than the linearscan and Chaitin-Briggs allocators on most benchmarks in a runtime compilation environment. By increasing allocation efficiency and preserving optimization quality, the presented algorithm increases the suitability and profitability of a graphcoloring register allocation strategy for a runtime compiler.