Communication and Computation Aware Task Scheduling Framework Toward Exascale Computing

Abstract

The race for Exascale Computing has naturally led computer architecture to transit from the multicore era and into the heterogeneous era. Exascale Computing within the heterogenous environment necessarily use the best-fit scheduling and resource utilization improvement. Task scheduling is the main critical aspect in managing the challenges of Exascale in the heterogenous computing environment. In this paper, a Communication and Computation Aware task scheduler framework (CCATSF) is introduced. The CCATSF framework consists of four parts; the first of which is the resource monitor, the second is the resources manager, the third is the task scheduler and the fourth is the dispatcher. The framework is based on a new hybrid task scheduling algorithm for a heterogenous computing environment. Our results are based on the random job generator that we implemented, and they indicate that the CCATSF framework, based on the proposed dynamic variant heterogenous early finish time (DVR-HEFT) algorithm is able to reduce the scheduler's makespan and increase the efficiency without increasing the algorithm's time complicity.

CQGA-HEFT: Q-learning-based DAG Scheduling Algorithm Using Genetic Algorithm in Clustered Many-core Platform

“…However, the self-driving system considered in this paper is modeled as a single DAG because dependencies exist between each function (e.g., sensing, localization, and path planning), and the final output must be produced by a defined deadline. DVR-HEFT [30] has been proposed by Sandokiji et al to optimize the performance and power consumption of heterogeneous resources. DVR-HEFT consists of a static list scheduling part and a dynamic scheduling part.…”