Providing fairness on shared-memory multiprocessors via process scheduling

Abstract

Competition for shared memory resources on multiprocessors is the most dominant cause for slowing down applications and makes their performance varies unpredictably. It exacerbates the need for Quality of Service (QoS) on such systems. In this paper, we propose a fair-progress process scheduling (FPS) policy to improve system fairness. Its strategy is to force the equallyweighted applications to have the same amount of slowdown when they run concurrently. The basic approach is to monitor the progress of all applications at runtime. When we find an application suffered more slowdown and accumulated less effective work than others, we allocate more CPU time to give it a better parity. Our policy also allows different weights to different threads, and provides an effective and robust tuner that allows the OS to freely make tradeoffs between system fairness and higher throughput.Evaluation results show that FPS can significantly improve system fairness by an average of 53.5% and 65.0% on a 4-core processor with a private cache and a 4-core processor with a shared cache, respectively. The penalty is about 1.1% and 1.6% of the system throughput. For memory-intensive workloads, FPS also improves system fairness by an average of 45.2% and 21.1% on 4-core and 8-core system respectively at the expense of a throughput loss of about 2%. 1

Addressing Fairness in SMT Multicores with a Progress-Aware Scheduler

“…Recent work has used the unfairness metric [12], [13], [14] for this purpose. This metric is defined as the maximum slowdown divided by the lowest slowdown across all the processes (N) of the workload as shown in Equation 2.…”

“…Therefore, other authors address aims to provide system fairness, by focusing on process scheduling. Fairness oriented process schedulers have been proposed by Fedorova et al [22] and Xu et al [13]. In [22], Fedorova et al target shared-cache contention using resource performance.…”

Contention-Aware Scheduling for SMT Multicore Processors

“…To overcome these problems, we decide to equalize the execution time of the benchmarks when running alone [11,19]. Hence, we measure the number of instructions that each benchmark completes when running alone in the system during x seconds 2 .…”

“…Other works try to improve system fairness by focusing on process scheduling. For instance, Xu et al [19] mainly target main memory bandwidth contention and propose a process scheduler that monitors the progress of the processes at run-time, which is used to increase the priority of the processes progressing at a slower pace.…”

“…Note that in the context of process scheduling, the turnaround time of a process is the time since the process is launched until it concludes its execution, including the time where the process is not scheduled [19]. Otherwise, by pausing threads contention could be reduced and the turnaround time of the individual processes improved.…”