63,153 Hits in 5.3 sec

A Load Balancing Tool for Distributed Parallel Loops

Ricolindo L. Cariño, Ioana Banicescu
2005 Cluster Computing  
This paper describes a tool for load balancing parallel loops on distributed-memory systems.  ...  The tool is useful for parallelizing sequential applications with parallel loops, or as an alternate load balancing routine for existing parallel applications.  ...  The load balancing tool described in this paper may be used to parallelize sequential applications with parallel loops, or as an alternate load balancing strategy for existing parallel applications.  ... 
doi:10.1007/s10586-005-4098-y fatcat:frr2dm4g4nauhpni6cuhppbcxu

Dynamic Load-Balancing for the STEM-II Air Quality Model [chapter]

J. Carlos Mouriño, María J. Martín, Patricia González, Ramón Doallo
2006 Lecture Notes in Computer Science  
The aim of this work is to improve load balance of the MPI parallel version of the STEM-II air quality model.  ...  Although all the data distributions presented here have been developed to be used with the STEM-II air quality model, they are also very suitable for use in other parallel applications.  ...  The code was initially parallelized using a static load balance strategy that turns out to be quite inefficient for some input data.  ... 
doi:10.1007/11751540_74 fatcat:qcdliir74jexdpbuk7cnhwxcii

Simulation of a hybrid model for image denoising

R.L. Carino, I. Banicescu, H. Lim, N. Williams, Seongjai Kim
2006 Proceedings 20th IEEE International Parallel & Distributed Processing Symposium  
The parallel simulation code was easily composed from existing serial code and a dynamic load balancing tool.  ...  We propose a new model for image denoising which is a hybrid of the total variation model and the Laplacian mean-curvature model.  ...  Dynamic Load Balancing Tool We describe in this section a dynamic load balancing tool we developed to simplify the parallelization and load balancing of applications that contain computationally-intensive  ... 
doi:10.1109/ipdps.2006.1639605 dblp:conf/ipps/CarinoBLWK06 fatcat:necs3lfg4rcd5k4pk6wunrojve

Detailed Load Balance Analysis of Large Scale Parallel Applications

Kevin A. Huck, Jesus Labarta
2010 2010 39th International Conference on Parallel Processing  
Balancing the workload in parallel applications is a difficult task, even in conventional cases. Many computing cycles are wasted when the load is not evenly balanced across processing nodes.  ...  In this paper, we consider the load balance of parallel applications which present unique challenges in the analysis process.  ...  Nearly all parallel performance analysis tools and methods include at least some rudimentary method for measuring and analyzing load balance.  ... 
doi:10.1109/icpp.2010.61 dblp:conf/icpp/HuckL10 fatcat:v4nlu2advrgbflbq2pudmgwfmy

Performance Study of Parallel Programming on Cloud Computing Environments Using MapReduce

Wen-Chung Shih, Shian-Shyong Tseng, Chao-Tung Yang
2010 2010 International Conference on Information Science and Applications  
Divisible load applications have such a rich source of parallelism that their parallelization can significantly reduce their total completion time on cloud computing environments.  ...  However, it is a challenge for cloud users, probably scientists and engineers, to develop their applications which can exploit the computing power of the cloud.  ...  With this metric, the program can distribute appropriate workloads to each node, and load balance can be achieved. The more accurate the estimation is, the better the load balance is.  ... 
doi:10.1109/icisa.2010.5480515 fatcat:umjfoxmhxbbnxc2coyauzsofui

Scalable load-balance measurement for SPMD codes

Todd Gamblin, Bronis R. de Supinski, Martin Schulz, Rob Fowler, Daniel A. Reed
2008 2008 SC - International Conference for High Performance Computing, Networking, Storage and Analysis  
We present and evaluate a novel technique for scalable, low-error load balance measurement. This uses a parallel wavelet transform and other parallel encoding methods.  ...  The overhead is low enough for online use in a production environment.  ...  INTRODUCTION In large-scale, distributed-memory parallel computers, balanced load is critical for system performance.  ... 
doi:10.1109/sc.2008.5222553 dblp:conf/sc/GamblinSSFR08 fatcat:os7rhbvvmvaqlavku7dodu7qhm

Optimizing load balance and communication on parallel computers with distributed shared memory [chapter]

Rudolf Berrendorf
1998 Lecture Notes in Computer Science  
To optimize programs for parallel computers with distributed shared memory two main problems need to be solved: load balance between the processors and minimization of interprocessor communication.  ...  This article describes a new technique called data-driven scheduling which can be used on sequentially iterated program regions on parallel computers with a distributed shared memory.  ...  I would like to thank the developers of the graph partitioning libraries I used in my work, namely: George Karypis (Metis), Chris Walshaw (Jostle), Bruce A.  ... 
doi:10.1007/bfb0057866 fatcat:p5n74jasijcl5cf76nnmzdrtqu

Controlling application grain size on a network of workstations

Bruce S. Siegell, Peter A. Steenkiste
1995 Proceedings of the 1995 ACM/IEEE conference on Supercomputing (CDROM) - Supercomputing '95  
Keywords Automatic parallelization, grain size, network of workstations, dynamic load balancing.  ...  In this paper we present a method for automatically selecting the grain size of the computation consisting of nested DO loops.  ...  This paper investigates the impact of grain size on parallelism, communication costs, and load balancing on a network of workstations, and presents methods for selecting and controlling grain size for  ... 
doi:10.1145/224170.224497 dblp:conf/sc/SiegellS95 fatcat:hluqyxnk6fdgpcxpa45qzjr6zq

Hierarchical Dynamic Loop Self-Scheduling on Distributed-Memory Systems Using an MPI+MPI Approach [article]

Ahmed Eleliemy, Florina M. Ciorba
2019 arXiv   pre-print
Such loops are often irregular and a balanced execution of their loop iterations is critical for achieving high performance.  ...  On distributed-memory systems, DLS techniques can be implemented using a hierarchical master-worker execution model and are, therefore, called hierarchical DLS techniques.  ...  Advanced Scientific Computing (PASC) project SPH-EXA: Optimizing Smooth Particle Hydrodynamics for Exascale Computing.  ... 
arXiv:1903.09510v1 fatcat:uysbknuuonhepmnu55vpl7fx7y

Fast data parallel polygon rendering

F. A. Ortega, C. D. Hansen, J. P. Ahrens
1993 Proceedings of the 1993 ACM/IEEE conference on Supercomputing - Supercomputing '93  
Issues involving load balancing are addressed a n d a d a t a p arallel load balancing algorithm is presented.  ...  This paper describes a data parallel method f o r p olygon rendering on a massively parallel machine.  ...  A load balancing algorithm was developed which balances this type of data parallel loop. The algorithm works only for data parallel loops with independent loop iterations.  ... 
doi:10.1145/169627.169820 dblp:conf/sc/OrtegaHA93 fatcat:7zbcypqnvjaxpiunjjxqo7kn54

An Approach for Realistically Simulating the Performance of Scientific Applications on High Performance Computing Systems [article]

Ali Mohammed, Ahmed Eleliemy, Florina M. Ciorba, Franziska Kasielke, Ioana Banicescu
2019 arXiv   pre-print
Dynamic loop self-scheduling (DLS) techniques are instrumental in improving the performance of scientific applications on HPC systems via load balancing.  ...  Scientific applications often contain large, computationally-intensive, and irregular parallel loops or tasks that exhibit stochastic characteristics.  ...  The DLB tool is an MPI-based dynamic load balancing library [23] .  ... 
arXiv:1910.06844v1 fatcat:gduoftfhgvhq5e3egnyva3zwqm

Toward Automatic Data Distribution for Migrating Computations

Lei Pan, Jingling Xie, Ming Kin Lai, Michael B. Dillencourt, Lubomir F. Bic
2007 Proceedings of the International Conference on Parallel Processing  
This paper focuses on data distribution for NavP.  ...  We introduce the Navigational Trace Graph (NTG), a mathematical structure that captures the alignment and distribution preferences of a sequential program.  ...  As mentioned earlier, our data distribution for DSC guarantees data load balancing but not necessarily computation load balancing.  ... 
doi:10.1109/icpp.2007.80 dblp:conf/icpp/PanXLDB07 fatcat:wl4g2q3xgfgfng5y3qyew2nwoe

Grain graphs

Ananya Muddukrishna, Peter A. Jonsson, Artur Podobas, Mats Brorsson
2016 Proceedings of the 21st ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming - PPoPP '16  
Problems with OpenMP parallel for-loops are similarly difficult to resolve since tools only visualize aggregate thread-level statistics such as load imbalance without zooming into a per-chunk granularity  ...  We present grain graphs, a new OpenMP performance analysis method that visualizes grains -computation performed by a task or a parallel for-loop chunk instance -and highlights problems such as low parallelism  ...  There is a similar problem with parallel for-loops where iterations are distributed to different threads in chunks.  ... 
doi:10.1145/2851141.2851156 dblp:conf/ppopp/MuddukrishnaJPB16 fatcat:qotlvbdykbeuhhylhajuxvwave

A parameter study of a hybrid Laplacian mean-curvature flow denoising model

Ioana Banicescu, Hyeona Lim, Ricolindo L. Cariño, Seongjai Kim
2010 Journal of Supercomputing  
This article presents results of a parameter study for a new denoising model, using parallel computing and advanced dynamic load balancing techniques for performance improvement of implementations.  ...  The hybrid model is simulated on a general-purpose Linux cluster for which the parallel efficiency exceeds 96%.  ...  For the parameter study, we used a dynamic load balancing Dynamic load balancing tool We describe in this section a dynamic load balancing tool we developed to simplify the parallelization and load  ... 
doi:10.1007/s11227-010-0417-z fatcat:nzuhzgswwnarlip2ern2e2gwcm

JavaSymphony, a Programming Model for the Grid [chapter]

Alexandru Jugravu, Thomas Fahringer
2004 Lecture Notes in Computer Science  
In previous work, JavaSymphony has been introduced as a high level programming model for performance-oriented distributed and parallel Java programs.  ...  We have extended JavaSymphony to simplify the development of Grid applications written in Java, allowing the programmer to control parallelism, load balancing, and locality at a high level of abstraction  ...  In this paper we describe important JavaSymphony mechanisms for controlling parallelism, load balancing, and locality, which are crucial for the Grid.  ... 
doi:10.1007/978-3-540-24688-6_4 fatcat:rxkgbd4lx5eu5ibmj2vqdsyobi
« Previous Showing results 1 — 15 out of 63,153 results