CoreTSAR: Adaptive Worksharing for Heterogeneous Systems

Abstract. The popularity of heterogeneous computing continues to in-crease rapidly due to the high peak performance, favorable energy e-ciency, and comparatively low cost of accelerators. However, heteroge-neous programming models still lack the flexibility of their CPU-only counterparts. Accelerated OpenMP models, including OpenMP 4.0 and OpenACC, ease the migration code from CPUs to GPUs but lack much of OpenMP’s flexibility: OpenMP applications can run on any num-ber of CPUs without extra user e↵ort, but GPU implementations do not o↵er similar adaptive worksharing across GPUs in a node, nor do they employ a mix of CPUs and GPUs. To address these shortcomings, we present CoreTSAR, our library for scheduling cores via a task-size adapting runtime system by supporting worksharing of loop nests across arbitrary heterogeneous resources. Beyond scheduling the computational load across devices, CoreTSAR includes a memory-management system that operates based on task association, enabling the runtime to dynam-ically manage memory movement and task granularity. Our evaluation shows that CoreTSAR can provide nearly linear scaling to four GPUs and all cores in a node without modifying the code within the parallel region. Furthermore, CoreTSAR provides portable performance across a variety of system configurations.