EffiSha

• DOI: 10.1145/3018743.3018748
• Publication Date: 2017-01-27T18:41:04Z
• Authors: Guoyang Chen; Yue Zhao; Xipeng Shen; Huiyang Zhou
• Keywords: Human multitasking; Turnaround time

Modern GPUs are broadly adopted in many multitasking environments, including data centers and smartphones. However, the current support for scheduling of multiple GPU kernels (from different applications) is limited, forming a major barrier to meet practical needs. This work first time demonstrates that on existing GPUs, efficient preemptive possible even without special hardware support. Specifically, it presents EffiSha, pure software framework enables with very low overhead. The enabled scheduler offers flexible priorities, significant potential reducing average turnaround improving system overall throughput programs share modern GPU.