In this paper, a workflow scheduling approach, which consists of two algorithms, is proposed. A submitted workflow is first partitioned into subgraphs on the global Grid level by the graph partitioning algorithm accor...
详细信息
There is a growing set of sensor-driven latency-sensitive applications that are exceeding the capabilities of what can be provided through cloud services and are driving the adoption of distributed edge computing. We ...
详细信息
In this paper, we investigate the suitability of load scheduling algorithms to workload characteristics. With a set of workloads classified by the amount of allocated resources and their allocation times, we perform e...
详细信息
ISBN:
(纸本)3540219463
In this paper, we investigate the suitability of load scheduling algorithms to workload characteristics. With a set of workloads classified by the amount of allocated resources and their allocation times, we perform experiments with various load scheduling algorithms. As adaptive scheduling algorithms, ARR and ALC are introduced on top of the ALBM (adaptive load balancing and management) cluster system. We present the performance results with a number of classified workloads and a realistic workload, explaining the performance relationship between workloads and scheduling algorithms.
Computer resource management and scheduling algorithms have been exploited in order to run applications in an efficient and effective way. Since this area is heavily exploited and has a range of application domains, i...
详细信息
In order to meet the testing requirements of task scheduling algorithms, this paper proposes two algorithms for generating random adjacency matrixes. One algorithm is used for generating random adjacency matrixes repr...
详细信息
The data warehouse is a centralized repository for analyzing and storing huge amounts of data. In distributed data warehouse, data can be shared across multiple data repositories which belong to one or more organizati...
详细信息
Hyperparameter optimization (HPO), the process of searching for optimal hyperparameter configurations for a model, is becoming increasingly important with the rise of automated machine learning. Using HPO can be incre...
详细信息
This paper is concerned with concurrent execution of multiple process groups on heterogeneous computing systems in non-dedicated parallel environment. The heterogeneity of the computing system is characterized by the ...
详细信息
In [5], we have proposed a compile-time optimization approach, the bottom-up top-down duplication heuristic (BTDH), for static scheduling of directed-acyclic graphs (DAGs) on distributed memory multiprocessors (DMMs)....
详细信息
ISBN:
(纸本)0818626305
In [5], we have proposed a compile-time optimization approach, the bottom-up top-down duplication heuristic (BTDH), for static scheduling of directed-acyclic graphs (DAGs) on distributed memory multiprocessors (DMMs). In this paper, we discuss the applications of BTDH for list scheduling algorithms (LSAs). There are two ways to use BTDH for LS As. (1) BTDH can be used with a LSA to form a new scheduling algorithm (LS A/BTDH). (2) Itcan be used as apure optimization algorithm for a LSA (LSA-BTDH). We have applied BTDH with two well known LSAs, the highest level first with estimated time (HLFET) and the earlier task first (ETF) heuristics. We have performed extensive simulation to study the performance of BTDH for LSAs. Three parameters, graph parallelism (GP) of a DAG [19], the ratio of average communication cost to average computation cost (CCR) of a DAG [5], and the processor number (PN) of a DMM, are simulated. From the simulation, we have the following conclusions. (I) Given a DAG, the GP of the DAG can accurately predict the number of processors to be used such that a good scheduling length and a good resource utilization (or efficiency) can be achieved simultaneously. (II) In terms of speedups, we have LS A/BTDH > LSA-BTDH> ETF > HLFET. Experimental results of scheduling FFT programs, which are written in Single Program Multiple Data (SPMD) programming approach, on NCUBE-2 are also presented. The results confirm our simulation results and show that the speedups of LS A/BTDH and LSA-BTDH are better than the speedups of LSAs.
In real-time systems, periodic tasks and aperiodic tasks exist simultaneously. In a uniprocessor system, mainly there are Deferrable Server algorithm (DS) [1], Slack Stealing algorithm (SSA) [2] and their extended ver...
详细信息
暂无评论