Compiler-Guided Throughput Scheduling for Many-core Machines

TL;DR

This paper introduces a compiler-guided, predictive scheduling framework for many-core systems that improves throughput by forecasting application behavior and proactively managing resources, outperforming traditional reactive schedulers.

Contribution

It presents a novel compiler-based approach that classifies loops, inserts beacons, and enables proactive scheduling to enhance throughput in multi-core environments.

Findings

Achieves an average throughput improvement of 76.78% over CFS.

Demonstrates up to 3.2x throughput increase on Amazon Graviton2.

Effectively predicts application behavior for better resource management.

Abstract

Typical schedulers in multi-tenancy environments make use of reactive, feedback-oriented mechanisms based on performance counters to avoid resource contention but suffer from detection lag and loss of performance. In this paper, we address these limitations by exploring the utility of predictive analysis through dynamic forecasting of applications' resource-heavy regions during its execution. Our compiler framework classifies loops in programs and leverages traditional compiler analysis along with learning mechanisms to quantify their behaviour. Based on the predictability of their execution time, it then inserts different types of beacons at their entry/exit points. The information produced by beacons in multiple processes is aggregated and analyzed by the proactive scheduler to respond to the anticipated workload requirements. For throughput environments, we develop a framework that demonstrates high-quality predictions and improvements in throughput over CFS by 76.78% on an average and up to 3.2x on Amazon Graviton2 Machine on consolidated workloads across 45 benchmarks.