Load balanced reliable task scheduling algorithm for heterogeneous systems

Dynamic Variant Rank HEFT Task Scheduling Algorithm Toward Exascle Computing

In this paper, aiming at a high miss deadline ratio in the large-scale task scheduling of heterogeneous parallel computing environment, the Arrived-Heterogeneous Earliest Finish Time (A-HEFT) algorithm based on the valid main path was proposed. This algorithm not only ensures all subtasks effectively processed within the time limit, but performs better in scheduling random arrival Directed Acyclic Graph (DAG) and the resource utilization among them during the execution period. Firstly, we proposed a new model of tasks produced by the same data source with a certain order based on the traditional model of DAG. Then, change the calculation of the upward rank of tasks in Heterogeneous Earliest Finish Time (HEFT) according to the task attribute of DAG, and get the execution priority from upward rank of sub-tasks. Finally, execute stretching scheduling based on idle time on the main path. Compared with Limited Preemptive Earliest Deadline First (LPEDF) algorithm and HEFT algorithm, the results of our work shows that A-HEFT algorithm performs better in processor utilization and success rate of task scheduling in heterogeneous system scheduling.

Computer vision applications have stringent performance constraints that must be satisfied when they are run at the edge on programmable low-power embedded devices. OpenVX has emerged as the de-facto reference standard to develop such applications. OpenVX uses a primitive-based programming model that results in a directed-acyclic graph (DAG) representation of the application, which can then be used for automatic system-level optimizations and synthesis to heterogeneous multi- and many-core platforms. Although OpenVX has been standardized, its state-of-the-art algorithm for task mapping and scheduling does not deliver the performance necessary for such applications to be deployed on heterogeneous multi-/many-core platforms. This article focuses on addressing this challenge with three main contributions: First, we implemented a static task scheduling and mapping approach for OpenVX using the heterogeneous earliest finish time (HEFT) heuristic. We show that HEFT allows us to improve the system performance up to 70 percent on one of the most widespread smart systems for applying computer vision and intelligent video analytics in general at the edge (i.e., NVIDIA VisionWorks on NVIDIA Jetson TX2). Second, we show that HEFT, in the context of a vision application for edge computing where some primitives may have multiple implementations (e.g., for CPU and GPU), can lead to load imbalance amongst heterogeneous computing elements (CEs), thus suffering from degraded performance. Third, we present an algorithm called exclusive earliest finish time (XEFT) that introduces the notion of exclusive overlap between single implementation primitives to improve the load balancing. We show that XEFT can further improve the system performance up to 33 percent over HEFT, and 82 percent over the native OpenVX scheduler. We present the results on a large set of benchmarks, including a real-world localization and mapping application (ORB-SLAM) combined with an NVIDIA inference application based on convolutional neural networks (CNNs) for object detection.

Task scheduling is a major issue in heterogeneous resource environments and particularly in multiprocessor system and Cloud. It means selection of the best suitable resources for task execution. Among the scheduling heuristics proposed in the literature, the Heterogeneous Earliest Finish Time (HEFT) heuristic is known by giving a good schedules in short time. In this paper, we propose three improvement versions of HEFT, where generally if certain conditions were verified, decisions made by the heuristic do not accounting a task being in scheduling only, but also take into account the impact of this decision to the children of the task. Thus, the heuristic can allocate this task and one of its children on the processor that can minimize execution time of the two tasks at the same time. Simulation results indicate that the new versions of HEFT can effectively reduce the schedule length in most cases without making the execution time prohibitively high otherwise.

Green cloud environment by using robust planning algorithm

Finding an optimal solution to the problem of scheduling an application modeled by a directed acyclic graph (DAG) onto a distributed system is known to be NP-complete. The complexity of the problem increases when task scheduling is to be done in a heterogeneous computing system, where the processors in the network may not be identical and take different amounts of time to execute the same task. This paper introduces a performance effective task scheduling (PETS) algorithm for network of heterogeneous system, with complexity O(v+e) (p+log v), which provides optimal results for applications represented by DAGs. The performance of the algorithm is illustrated by comparing the schedule length, speedup, efficiency and the scheduling time with existing algorithms such as, heterogeneous earliest finish time (HEFT) and critical-path on a processor (CPOP) and levelized min time (LMT) reported in this paper. The comparison study based on both randomly generated graphs and graphs of some real applications shows that PETS algorithm substantially outperforms existing algorithms

In recent years, Artificial Neural Networks have evolved rapidly and are applied to various fields. Meanwhile, to enhance computation efficiency of neural network applications, more and more neural network accelerators have been developed. Though traditional task scheduling algorithms on heterogeneous systems have been intensively researched, they can’t be applied to neural network accelerators directly. Based on typical characteristics of neural network accelerators, we formalize the problem of tasks scheduling for neural networks, and transplant two listing heuristic scheduling algorithms, Heterogeneous-Earliest-Finish-Time (HEFT) and Critical-Path-on-a-Processor (CPOP). Inspired by the separable features of neural network operations, we propose two partition algorithms, the Iterative Partition Scheduling Algorithm (IPS) and the Partition Scheduling Combination Algorithm (PSC), which can be associated with scheduling algorithms. Further, we conduct experiments on some typical neural networks, and results show that compared to scheduling-only algorithms the partition associated algorithms achieve about 2x to 3x speedup.

In the rapidly evolving landscape of cloud computing, scheduling complex scientific workflows poses significant challenges, particularly with constraints like budget considerations. While recent years have seen considerable research focus on budget-constrained cloud workflow scheduling, existing studies primarily concentrate on constructing solutions without delving into potential enhancements. To address this gap, our paper introduces PACP-HEFT, a modified HEFT algorithm integrating priority adjustment and task optimization to minimize makespan. Leveraging the Heterogeneous Earliest-Finish-Time (HEFT) algorithm as its foundation, PACP-HEFT incorporates two priority adjusters informed by task characteristics gleaned from task dependency topology and real-time scheduling data. Additionally, a critical task optimizer conducts thorough analyses and optimizations to reduce overall makespan. Extensive experimentation with real-world workflows underscores PACP-HEFT’s superior performance compared to contemporary algorithms. Note to Practitioners—This paper presents a deterministic approach to budget-constrained scheduling problem, with the objective of minimizing the makespan. The proposed PACP-HEFT algorithm exhibits robust solution enhancement capabilities through the utilization of two priority adjusters and a critical task optimizer. The two priority adjusters fine-tune task priorities based on characteristics extracted from task dependency topology and real-time scheduling data, while the critical task optimizer performs an in-depth analysis of scheduling solutions and optimizes critical tasks to reduce overall makespan. Real-world workflow experiments clearly highlight the superior performance of PACP-HEFT compared to state-of-the-art algorithms. Validation experiments also underscore the effectiveness of each priority adjuster and the critical task optimizer. Moreover, PACP-HEFT’s ability to rapidly reach solutions further enhances its practicality, ensuring timely scheduling outcomes. Furthermore, we assess the adaptability and transferability of these components, emphasizing their potential applicability in enhancing the performance of other scheduling algorithms. For practical applications, this algorithm can be directly applied to rapidly minimize the makespan of budget-constrained scheduling problems in the cloud. Moreover, it can be integrated into other schedulers as a solution enhancement component, offering versatility and improved scheduling efficiency.

Energy-efficient triple modular redundancy scheduling on heterogeneous multi-core real-time systems

Performance-, power-, and energy-aware scheduling techniques play an essential role in optimally utilizing processing elements (PEs) of heterogeneous systems. List schedulers, a class of low-complexity static schedulers, have commonly been used in static execution scenarios. However, list schedulers are not suitable for runtime decision making, particularly when multiple concurrent applications are interleaved dynamically. For such cases, the static task execution times and expectation of idle PEs assumed by list schedulers lead to inefficient system utilization and poor performance. To address this problem, we present techniques for optimizing execution of list scheduling algorithms in dynamic runtime scenarios via a family of algorithms inspired by the well-known heterogeneous earliest finish time (HEFT) list scheduler. Through dynamically arriving, realistic workload scenarios that are simulated in an open-source discrete event heterogeneous SoC simulator, we exhaustively evaluate each of the proposed algorithms across two SoCs modeled after the Xilinx Zynq Ultrascale+ ZCU102 and O-Droid XU3 development boards. Altogether, depending on the chosen variant in this family of algorithms, we are able to achieve an up to 39% execution time improvement, up to 7.24x algorithmic speedup, or up to 30% energy consumption improvement compared to the baseline HEFT implementation.

A fast predictive algorithm with idle reduction for heterogeneous system scheduling

Traditional human-designed heuristics-based algorithms are commonly employed to address workflow scheduling problems. Such as Heterogeneous Earliest Finish Time (HEFT) and CriticalPath are two best-known list based heuristics. Generally, heuristics-based approaches can generate a single heuristic by making decisions based on the current status of the tasks and available resources and map the unscheduled tasks to the available resources. However, traditional heuristics can easily cause unbalancing load problem. For example, with the objective of minimizing the makespan, traditional heuristics prefer to use computation resources with high computation capacity. Usually, such resources are expensive, which will lead to a high cost as a result. Therefore it is hard for single traditional heuristics to do a trade-off; thus, traditional heuristics are hard to apply for multi-objective workflow scheduling. In this paper, we develop a novel algorithm for workflow scheduling by considering both minimizing cost and makespan. Experiments show that the MOSGP approach can effectively minimize makespan and cost simultaneously.

Efficient scheduling algorithm is one of the key factors in determining the performance of distributed memory machines. This paper presents an enhanced Search and Du plication Based Scheduling (SDBS) algorithm which car schedule Directed Acyclic Graphs (DA Gs) with variable execution time tasks. Since the DAGs are based on data dependencies of the tasks, the control dependencies are clustered within each node. If the node encapsulates a loop or an if-then-else kind of statement, then assuming the execution times to be fixed would not be appropriate. It is assumed in this paper that the execution time varies between two time estimates. The complexity of this scheduling algorithm is in 0(V 4- E), where V is the number of node,- and E is the number of edges in the task graph. This algo rithm is based on some realistic assumptions and generates an optimal time schedule. If the assumptions cannot be completely satisfied then the algorithm provides a schedule which is close to optimal. The performance in these cases has been obtained using extensive simulation work which indicate the closeness of the results.

As the large amount of video surveillance data floods into cloud data center, achieving load balancing in a cloud network has become a challenging problem. Meanwhile, we hope the cloud data center maintains low latency, low consumption, and high throughput performance when transmitting massive amounts of data. OpenFlow enables a software-defined solution through programing to control the scheduling of data flow in the cloud data center. However, the existing scheduling algorithm of the data center cannot cope with the congestion of the network center effectively. Even for some dynamic scheduling algorithms, adjustments can only be made after congestion occurs. Hence, we propose a proactive and dynamically adjusted mixed-flow load-balanced scheduling (MFLBS) algorithm, which not only takes into account the different sizes of flows in the network but also maintains maximum throughput while balancing the load. In this paper, the MFLBS problem was formulated, along with a set of heuristic algorithms for real-time feedback and adjustment. Experiments with mesh and tree network models show that our MFLBS is significantly better than other dynamic scheduling algorithms, including one-hop DLBS and static scheduling algorithm FCFS. The MFLBS algorithm can effectively reduce the delay of small flows and average delay while maintaining high throughput.

Enhanced wireless communication improves the connectivity of vehicular networks in which vehicles are utilized as infrastructures for communication and computation. Thus, a new concept “Vehicular Edge Computing (VEC)” is formed. As VEC utilizes a collaborative multitude of near-user edge resources (i.e. vehicles) in the Internet of Vehicles, the capability of these joint resources becomes heterogeneous especially in their movements. Therefore, one critical problem is how to efficiently schedule each task under such mobile environments. For the reason, we propose a hybrid dynamic scheduling scheme (HDSS) that has the ability to optimize the task scheduling dynamically based on the changeable system environments. HDSS provides a decision function (DF) to select a better-performed scheduling algorithm from two provided candidates: the queue-based dynamic scheduling (QDS) algorithm and the time-based dynamic scheduling (TDS). QDS coincides with the Join-the-Shortest Queue scheme, which decides the scheduling by sorting out a server with the shortest queue-length; nevertheless, TDS is novel scheme that is designed to implement task allocation by estimating the waiting time of each server in order to select a server with the fastest response. Finally, this research generates formal models of each scheduling algorithm and the hybrid scheduling scheme to conduct performance evaluation with a fluid flow approximation technique. The analysis results in a superior performance of HDSS in the unstable VEC environments.