This repository provides the source code used to generate all results presented in the paper "An efficient static scheduling heuristic for accelerated architectures," which introduces a new static task scheduling heuristic for accelerated architectures called Heterogeneous Optimistic Finish Time (HOFT).
To evaluate HOFT, we implemented a simple software framework which allows the simulated scheduling of user-defined task graphs (specifically, Directed Acyclic Graphs or DAGs) on user-defined CPU-GPU target platforms. The Python source code for this can be found in the folder simulator. Details on which Python packages are required and other installation information are provided in the folder's main README file.
The LaTeX source for the paper and all other relevant files are contained in the folder An_efficient_new_static_scheduling_heuristic_for_accelerated_architectures.
This project is licensed under the GPL-3.0 License - see LICENSE for details.
[1] T. McSweeney, N. Walton and M. Zounon. An efficient static scheduling heuristic for accelerated architectures. Submitted to the International Conference on Computational Science 2020 (https://www.iccs-meeting.org/iccs2020/).