GPGPU Scheduling Schemes to Improve Latency and Resource Utilization

Article Sidebar

Published: Apr 30, 2025

Main Article Content

Narayana Murty N*
Department of Electronics and Communication Engineering, Lovely Professional University, Phagwara, India
Dr.Harjit Singh
Department of Computer Science and Engineering, Lovely Professional University, Phagwara, India
Dr. Sukhmani K. Thethi
Department of Electronics and Communication Engineering, Lovely Professional University, Phagwara, India

Abstract

Real-time embedded systems today need GPU scheduling methods that are fast, flexible, and energy-saving while handling different types of tasks with limited resources. Current schedulers, like federated and hierarchical ones, often have problems giving real-time guarantees, adapting to changing needs, and saving power — especially on AMD GPUs that have limited Local Data Share (LDS) and High Bandwidth Memory (HBM). This paper presents a new Hybrid GPGPU Scheduler that solves these problems. It combines static, dynamic, and machine learning (ML)-based scheduling with task models that are aware of memory limits. Using Gradient Boosting Decision Trees (GBDT)/XGBoost, the scheduler predicts task features and decides if they can meet deadlines, while smartly assigning Compute Units (CUs).
Tests on a Linux platform with PoCL and ROCm show that the scheduler beats others like RTGPU, Self-Suspension, and Enhanced MPCP. It improves task scheduling by 11% at high usage (U = 2.0), accepts 85% of tasks when 12 run together, and cuts scheduling delay by 40%. It also saves 9% more energy per watt while keeping power use low (1.0–2.0 W).
Resource use stays high, with 92% CU and 85% CPU usage. These results prove that the Hybrid GPGPU Scheduler is a scalable and energy-efficient solution for real-time GPU scheduling in embedded systems, balancing performance, flexibility, and power savings in systems with tasks of different critical levels.

Article Details

Issue
Vol. 167 No. A2 (S) (2025): Technologies and Their Effects on Real-Time Social Development
Section
Articles

References

Z. Zou et al., “RTGPU: Real-time GPU scheduling of hard deadline parallel tasks with fine-grain utilization,” in Proc. IEEE Real-Time Syst. Symp., 2023, pp. 1–12.

Srinivasa Sai Abhijit Challapalli, “Optimizing Dallas-Fort Worth Bus Transportation System Using Any Logic,” Journal of Sensors, IoT & Health Sciences, vol.2, no.4, 40-55, 2024.

R. Nozal and J. L. Bosque, “EngineCL: Usability and performance in heterogeneous computing,” Future Gener. Comput. Syst., vol. 108, pp. 153–165, 2020.

Srinivasa Sai Abhijit Challapalli. “Sentiment Analysis of the Twitter Dataset for the Prediction of Sentiments,” Journal of Sensors, IoT & Health Sciences, vol.2, no.4, pp.1-15, 2024.

G. Campeanu et al., “Component-based development of embedded systems with GPUs,” in Proc. Euromicro Conf. Softw. Eng. Adv. Appl., 2020, pp. 1–8.

J. Lee and M. A. Al Faruque, “Spatial temporal scheduler,” in Proc. Des. Autom. Conf., 2016, pp. 1–6.

S. Hosseinimotlagh and Y. Kim, “Thermal-aware servers for real-time tasks on multi-core GPU-integrated embedded systems,” in Proc. IEEE Real-Time Embedded Technol. Appl. Symp., 2018, pp. 1–10.

N. Capodieci et al., “Deadline-based scheduling for GPU with preemption support,” in Proc. IEEE Real-Time Syst. Symp., 2018, pp. 1–12.

X. Zhou et al., “GPES: A preemptive execution system for GPGPU computing,” in Proc. IEEE Real-Time Embedded Technol. Appl. Symp., 2015, pp. 1–10.

B. Wu et al., “A model-based software solution for simultaneous multiple kernels on GPUs,” ACM Trans. Embedded Comput. Syst., vol. 19, no. 5, pp. 1–25, 2020.

S. Kato et al., “TimeGraph: GPU scheduling for real-time multi-tasking environments,” in Proc. USENIX Annu. Tech. Conf., 2011, pp. 1–14.

G. A. Elliott and J. H. Anderson, “Globally scheduled real-time multiprocessor systems with GPUs,” in Proc. Int. Conf. Real-Time Comput. Syst. Appl., 2011, pp. 1–10.

G. A. Elliott et al., “GPUSync: A framework for real-time GPU management,” in Proc. IEEE Real-Time Syst. Symp., 2013, pp. 1–12.

S. Kang et al., “Priority-driven spatial resource sharing (PR-SRS),” in Proc. IEEE Real-Time Embedded Technol. Appl. Symp., 2017, pp. 1–10.

H. Choi et al., “An efficient scheduling scheme using estimated execution time,” in Proc. IEEE Int. Conf. Embedded Real-Time Comput. Syst. Appl., 2013, pp. 1–8.

V. Raca and E. Mehofer, “ClusterCL: Comprehensive support for multi-kernel data-parallel applications,” in Proc. Int. Conf. High Perform. Comput., 2020, pp. 1–10.

A. M. Aji et al., “MultiCL: Enabling automatic scheduling for task-parallel workloads,” in Proc. Int. Conf. Supercomput., 2016, pp. 1–12.

J. Kim and Y. Kim, “Interference-aware execution framework with Co-scheML on GPU clusters,” in Proc. IEEE Int. Conf. Cluster Comput., 2023, pp. 1–10.

Y. Wang et al., “Dynamic GPU scheduling with multi-resource awareness and live migration support,” in Proc. IEEE Int. Conf. Mach. Learn. Appl., 2023, pp. 1–8.

R. Nozal and J. L. Bosque, “Straightforward heterogeneous computing with the oneAPI coexecutor runtime,” in Proc. Int. Conf. Parallel Distrib. Comput., 2021, pp. 1–10.

M. A. Dávila Guzmán et al., “Cooperative CPU, GPU, and FPGA heterogeneous execution with EngineCL,” in Proc. Int. Conf. High Perform. Comput., 2019, pp. 1–10.

J. Zhao et al., “ISPA: Exploiting intra-SM parallelism in GPUs via fine-grained resource management,” in Proc. Int. Conf. Supercomput., 2023, pp. 1–12.

G. Yu et al., “SMGuard: A flexible and fine-grained resource management framework for GPUs,” in Proc. IEEE Int. Symp. High-Perform. Comput. Archit., 2018, pp. 1–12.

E. Barreiros and A. Melo, “Efficient microscopy image analysis on CPU-GPU systems with cost-aware irregular data partitioning,” in Proc. Int. Conf. Image Process., 2022, pp. 1–8.

A. Elsabbagh et al., “Vortex: OpenCL compatible RISC-V GPGPU,” in Proc. IEEE Int. Symp. Field-Program. Custom Comput. Mach., 2020, pp. 1–8.

K. Suzuki et al., “GPUvm: GPU virtualization at the hypervisor,” in Proc. USENIX Annu. Tech. Conf., 2016, pp. 1–14.