Fine-grained concurrent kernel execution on SM/CU level for GPGPU computing

Abstract

As a critical computing resource in multi-user systems such as supercomputers, data centres and cloud services, a GPU contains multiple compute units (CUs). GPU Multitasking is an intuitive solution to underutilization in GPGPU computing. Recently proposed solutions of multitasking GPUs can be classified into two categories: (1) spatially-partitioned sharing (SPS), which co-executes different kernels on disjointed sets of compute units (CU), and (2) simultaneous multikernel (SMK), which runs multiple kernels simultaneously within a CU. Compared to SPS, SMK can improve resource utilization even further due to the interleaving of instructions from kernels with low dynamic resource contentions. However, it is hard to implement SMK on current GPU architecture, because: (1) techniques for applying SMK on top of GPU hardware scheduling policy are scarce; (2) finding an efficient SMK scheme is difficult due to the complex interferences of concurrently executed kernels. In this paper, we propose a lightweight and effective performance model to evaluate the complex interferences of SMK. Based on the probability of independent events, our performance model is built from a totally new angle and contains limited parameters. Then, we propose a metric, symbiotic factor, which can evaluate an SMK scheme so that kernels with complementary resource utilization can co-run within a CU. Also, we analyse the advantages and disadvantages of kernel slicing and kernel stretching techniques and integrate them to apply SMK on GPUs instead of simulators. We validate our model on 18 benchmarks. Compared to the optimized hardware-based concurrent kernel execution whose kernel launching order brings fast execution time, the results of co-running kernel pairs show 11%, 18% and 12% speedup on AMD R9 290X, RX 480 and Vega 64, on average. Compared to the Warped-Slicer, the results show 29%, 18% and 51% speedup on AMD R9 290X, RX 480 and Vega 64, on average.