Singe: Leveraging warp specialization for high performance on GPUs

We present Singe, a Domain Specific Language (DSL) compiler for combustion chemistry that leverages warp specialization to pro-duce high performance code for GPUs. Instead of relying on tra-ditional GPU programming models that emphasize data-parallel computations, warp specialization allows compilers like Singe to partition computations into sub-computations which are then as-signed to different warps within a thread block. Fine-grain synchro-nization between warps is performed efficiently in hardware using producer-consumer named barriers. Partitioning computations us-ing warp specialization allows Singe to deal efficiently with the irregularity in both data access patterns and computation. Further-more, warp-specialized partitioning of computations allows Singe to fit extremely large working sets into on-chip memories. Finally, we describe the architecture and general compilation techniques necessary for constructing a warp-specializing compiler. We show that the warp-specialized code emitted by Singe is up to 3.75X faster than previously optimized data-parallel GPU kernels