QoSMT: supporting precise performance control for simultaneous multithreading architecture

Abstract

Simultaneous multithreading (SMT) technology improves CPU throughput, but also causes unpredictable performance fluctuations for co-located workloads. Although recent major SMT processors have adopted some techniques to promote hardware support for quality-of-service (QoS), achieving both precise performance control and high throughput on SMT architectures is still a challenging open problem. In this paper, we perform some comprehensive experiments on real SMT systems and cycle-accurate simulators. From these experiments, we observe that almost all in-core resources may suffer from severe contention as workloads vary. We consider this observation as the fundamental reason leading to the challenging problem above. Thus, we introduce QoSMT, a novel hardware scheme that leverages a closed-loop controlling mechanism to enforce precise performance control for specific targets, e.g. achieving 85%, 90% or 95% of the performance of a workload running alone respectively. We implement a prototype on GEM5 simulator. Experimental results show that the control error is only 1.4%, 0.5% and 3.6%.

Publication
In Conference on Supercomputing

QoSMT是我2016年时和靳鑫师兄一起完成的工作,也是我本科毕业设计的主要内容,主体代码在:QoSMT project.

Zhou, Yaoyang
Zhou, Yaoyang
Architect of LLM DSA; Maintainer of u-arch simulator for Xiangshan; PhD of Computer Architecture

I specialize in LLM DSA and CPU micro-architecture.