A comparison study of static mapping heuristics for a class of meta-tasks on heterogeneous computing systems

Braun, Tracy D.; Siegal, H.J.; Beck, Noah; Bölöni, Ladislau; Maheswaran, Muthucumaru; Reuther, Albert; Robertson, J; Theys, Mitchell D.; Yao, Bin; Hensgen, D.; Freund, Richard F.

doi:10.1109/hcw.1999.765093

Cited by 182 publications

(134 citation statements)

References 18 publications

Supporting

Mentioning

133

Contrasting

Unclassified

Order By: Relevance

“…Given a function for each architecture that describes the run-time of a task on this processing unit they propose different heuristics to minimize the overall finishing time. In addition to those and other static techniques [6], several dynamic methods have been proposed [21,29,17].…”

Section: Related Workmentioning

confidence: 99%

A Static Task Partitioning Approach for Heterogeneous Systems Using OpenCL

Grewe

O’Boyle

2011

Lecture Notes in Computer Science

152

124

View full text Add to dashboard Cite

Abstract. Heterogeneous multi-core platforms are increasingly prevalent due to their perceived superior performance over homogeneous systems. The best performance, however, can only be achieved if tasks are accurately mapped to the right processors. OpenCL programs can be partitioned to take advantage of all the available processors in a system. However, finding the best partitioning for any heterogeneous system is difficult and depends on the hardware and software implementation.We propose a portable partitioning scheme for OpenCL programs on heterogeneous CPU-GPU systems. We develop a purely static approach based on predictive modelling and program features. When evaluated over a suite of 47 benchmarks, our model achieves a speedup of 1.57 over a state-of-the-art dynamic run-time approach, a speedup of 3.02 over a purely multi-core approach and 1.55 over the performance achieved by using just the GPU.

show abstract

Section: Related Workmentioning

confidence: 99%

A Static Task Partitioning Approach for Heterogeneous Systems Using OpenCL

Grewe

O’Boyle

2011

Lecture Notes in Computer Science

152

124

View full text Add to dashboard Cite

show abstract

“…Intuitively, it attempts to map as many tasks as possible to their first choice of machine. If more time is available for finding better mapping, a more complex heuristic such as GA and SA should be considered [79]. …”

Section: Braun's 1999 Modelmentioning

confidence: 99%

“…The problem of mapping large and diverse groups of tasks onto the machines of an HC had been researched by Siegel et al The mapping problem is formulated as follows [79]:…”

Section: Braun's 1999 Modelmentioning

confidence: 99%

Task mapping on supercomputers with cellular networks

Chen

2008

Computer Physics Communications

View full text Add to dashboard Cite

S S S t t t o o o n n n y y y B B B r r r o o o o o o k k k U U U n n n i i i v v v e e e r r r s s s i i i t t t y y yThe official electronic file of this thesis or dissertation is maintained by the University Libraries on behalf of The Graduate School at Stony Brook University. Stony Brook University 2008This thesis focuses on techniques of task mapping for solving problems on parallel computers with hundreds of thousands of processors on cellular networks. Task mapping is a serious intellectual challenge and a practical tool for unleashing the potential power of supercomputers. It is challenging because of both the astronomical searching space and the high dependence on the exact nature of the applications and the computers. In this thesis, we propose two general static mapping models to optimize the assignment of tasks on heterogeneous, distributed-memory, ultra-scalable computers. In our models, the underlying application problems can be appropriately decomposed to subtasks with known computational load and known inter-task communiiii cational demands. We also know, or can conveniently measure, the computing systems' specifications such as individual processor speed and inter-processor communication cost. Our models abstract an application as a demand matrix and a parallel computer as a load matrix and a supply matrix with which we construct our models as minimizing the objective function value for completing the application on the given computer.We have tested several applications on Blue Gene/L supercomputer with 3D mesh and torus networks. For a 2D wave equation, the mappings generated by our models reduced communication by 51% for 3D-mesh and 31% for 3D-torus over the default MPI rank order mapping. For SMG2000 application, our mapping can reduce communication and total time by 16% and 5% over the default MPI rank order mapping, respectively. For NPB MG, we improve the communication time and benchmark result by 53% and 13%, respectively.For NPB CG, we improve the communication time and benchmark result by 43% and 22%, respectively. We believe that our models are useful for task assignment for broad applications on a family of supercomputers with cellular networks.iv

show abstract

“…Although genetic algorithms provide good quality schedules, their execution times are significantly higher than other alternatives. Extensive tests are required to find optimal values for the set of control parameters used in GA-based solutions [14] .…”

Section: Introdutionmentioning

confidence: 99%

“…Because of its key importance on performance, the task scheduling problem in general has been extensively studied and various heuristics have been proposed in the literature [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] . These heuristics are classified into a variety of categories such as list scheduling algorithms, clustering algorithms, guided random search methods and task duplication based algorithms.…”

Section: Introdutionmentioning

confidence: 99%

Low Complexity Performance Effective Task Scheduling Algorithm for Heterogeneous Computing Environments

Ilavarasan¹,

Thambidurai²

2007

J. of Computer Science

149

View full text Add to dashboard Cite

Abstract:A heterogeneous computing environment is a suite of heterogeneous processors interconnected by high-speed networks, thereby promising high speed processing of computationally intensive applications with diverse computing needs. Scheduling of an application modeled by Directed Acyclic Graph (DAG) is a key issue when aiming at high performance in this kind of environment. The problem is generally addressed in terms of task scheduling, where tasks are the schedulable units of a program. The task scheduling problems have been shown to be NP-complete in general as well as several restricted cases. In this study we present a simple scheduling algorithm based on list scheduling, namely, low complexity Performance Effective Task Scheduling (PETS) algorithm for heterogeneous computing systems with complexity O (e) (p+ log v), which provides effective results for applications represented by DAGs. The analysis and experiments based on both randomly generated graphs and graphs of some real applications show that the PETS algorithm substantially outperforms the existing scheduling algorithms such as Heterogeneous Earliest Finish Time (HEFT), Critical-Path-On a Processor (CPOP) and Levelized Min Time (LMT), in terms of schedule length ratio, speedup, efficiency, running time and frequency of best results.Key words: DAG, task graph, task scheduling, heterogeneous computing system, schedule length, speedup, efficiency INTRODUTIONA growing emphasis on concurrent processing of jobs has lead to an increased acceptance of heterogeneous computing environments and the availability of a network of processors makes a costeffective utilization of underlying parallelism for applications like weather modeling, image processing, real-time and distributed database systems. A wellknown strategy behind efficient execution of a huge application on a heterogeneous computing environment is to partition it into multiple independent tasks and schedule such tasks over a set of available processors. A task-partitioning algorithm takes care of efficiently dividing an application into tasks of appropriate grain size and an abstract model of such a partitioned application can be represented by a Directed Acyclic Graph (DAG). Each task of a DAG corresponds to a sequence of operations and a directed edge represents the precedence constraints between the tasks. Each task can be executed on a processor and the directed edge shows transfer of relevant data from one processor to another. Task scheduling can be performed at compile-time or at run-time. When the characteristics of an application, which includes execution times of tasks on different processors, the data size of the communication between tasks and the task dependencies, are known a priori, it is represented with a static model. The objective of task scheduling is to map the tasks on the processors and order their execution so that task precedence requirements are satisfied and a minimum overall completion time is obtained. The problem of scheduling of tasks with required precedence rela...

show abstract

A comparison study of static mapping heuristics for a class of meta-tasks on heterogeneous computing systems

Cited by 182 publications

References 18 publications

A Static Task Partitioning Approach for Heterogeneous Systems Using OpenCL

A Static Task Partitioning Approach for Heterogeneous Systems Using OpenCL

Task mapping on supercomputers with cellular networks

Low Complexity Performance Effective Task Scheduling Algorithm for Heterogeneous Computing Environments

Contact Info

Product

Resources

About