Filters








297 Hits in 7.4 sec

Using multiple energy gears in MPI programs on a power-scalable cluster

Vincent W. Freeh, David K. Lowenthal
2005 Proceedings of the tenth ACM SIGPLAN symposium on Principles and practice of parallel programming - PPoPP '05  
For example, IS using multiple gears uses 9% less energy and executes in 1% less time than the closest single-gear solution.  ...  Specifically, this paper shows that more than half of the NAS benchmarks exhibit a better energy-time tradeoff using multiple gears than using a single gear.  ...  Performance results were obtained on the NAS suite on a real power-scalable cluster composed of AMD Athlon-64 frequency and voltage scalable processors.  ... 
doi:10.1145/1065944.1065967 dblp:conf/ppopp/FreehL05 fatcat:w4gpiazizjbrvb37xiirhe5sga

Minimizing execution time in MPI programs on an energy-constrained, power-scalable cluster

Robert Springer, David K. Lowenthal, Barry Rountree, Vincent W. Freeh
2006 Proceedings of the eleventh ACM SIGPLAN symposium on Principles and practice of parallel programming - PPoPP '06  
The problem we address in this paper is: given a target program, a power-scalable cluster, and an upper limit for energy consumption, choose a schedule (number of nodes and CPU frequency) that simultaneously  ...  One way to reduce power (and therefore energy) requirements is to use high-performance cluster nodes that are frequency-and voltagescalable (e.g., AMD-64 processors).  ...  Hence, given a power-scalable cluster, there are two primary ways to save energy: (1) power down a subset of the nodes, and (2) on the nodes that are actively participating in the computation, scale down  ... 
doi:10.1145/1122971.1123006 dblp:conf/ppopp/SpringerLRF06 fatcat:vbbhtiu2efdl3lkw2vtgmass3e

Analyzing the Energy-Time Trade-Off in High-Performance Computing Applications

Vincent W. Freeh, David K. Lowenthal, Feng Pan, Nandini Kappiah, Rob Springer, Barry L. Rountree, Mark E. Femal
2007 IEEE Transactions on Parallel and Distributed Systems  
This paper analyzes the energy-time trade-off of a wide range of applications-serial and parallel-on a power-scalable cluster.  ...  We use a cluster of frequency and voltage-scalable AMD-64 nodes, each equipped with a power meter.  ...  We have also examined the energy saving potential by using multiple energy gears in MPI programs [23] .  ... 
doi:10.1109/tpds.2007.1026 fatcat:qlbutplhuvcedcyik5ol4nsd6u

Profile-based optimization of power performance by using dynamic voltage scaling on a PC cluster

Y. Hotta, M. Sato, H. Kimura, S. Matsuoka, T. Boku, D. Takahashi
2006 Proceedings 20th IEEE International Parallel & Distributed Processing Symposium  
With this system we examined the effectiveness of our optimization algorithm on two types of power-scalable clusters (Crusoe and Turion).  ...  In this paper, we propose a method of profiledbased power-performance optimization by DVS scheduling in a high-performance PC cluster.  ...  Acknowledgment This research is partly supported by the Japan Science and Technology Agency as a CREST research program entitled "Mega-Scale Computing Based on Low-Power Technology and Workload Modeling  ... 
doi:10.1109/ipdps.2006.1639597 dblp:conf/ipps/HottaSKMBT06 fatcat:5blzfcol7nbohizrstkc3idv7e

Just-in-time dynamic voltage scaling: Exploiting inter-node slack to save energy in MPI programs

Vincent W. Freeh, Nandini Kappiah, David K. Lowenthal, Tyler K. Bletsch
2008 Journal of Parallel and Distributed Computing  
On power-scalable clusters, one opportunity for saving energy with little or no loss of performance exists when the computational load is not perfectly balanced.  ...  As a result, interest in using powerscalable clusters, where frequency and voltage can be dynamically modified, has increased.  ...  A trivial improvement possible in a scalable cluster is to shift to the lowest gear before issuing the HALT.  ... 
doi:10.1016/j.jpdc.2008.04.007 fatcat:lubvujqwcrfbfd3pao66g3t33a

Sustainable GPU Computing at Scale

Justin Y. Shi, Moussa Taifi, Abdallah Khreishah, Jie Wu
2011 2011 14th IEEE International Conference on Computational Science and Engineering  
The idea is to leverage the powers of statistic multiplexing to tame the nagging HPC scalability challenges.  ...  Since energy consumption is defined by the number of components used, we consider a sustainable high performance computing (HPC) application can allow better performance and reliability at the same time  ...  This research is supported in part by a grant from the National Science Foundation CNS 0958854.  ... 
doi:10.1109/cse.2011.55 dblp:conf/cse/ShiTKW11 fatcat:7xtxfdioqncgtnjscuyqunhk7e

A feasibility analysis of power-awareness and energy minimization in modern interconnects for high-performance computing

Reza Zamani, Ahmad Afsahi, Ying Qian, Carl Hamacher
2007 2007 IEEE International Conference on Cluster Computing  
They discovered that more than half their applications incur less energy using multiple gears than using a single gear.  ...  In this work, we focus on the node/cluster power consumption. However, the CPU is a major power consumer in a node/cluster.  ... 
doi:10.1109/clustr.2007.4629224 dblp:conf/cluster/ZamaniAQH07 fatcat:bqfxcvt3tjdxvkx3w34c6eenum

Distributed TensorFlow with MPI [article]

Abhinav Vishnu, Charles Siegel, Jeffrey Daily
2017 arXiv   pre-print
In this paper, we extend recently proposed Google TensorFlow for execution on large scale clusters using Message Passing Interface (MPI).  ...  We evaluate our implementation using an InfiniBand cluster and several well knowndatasets. Our evaluation indicates the efficiency of our proposed implementation.  ...  MPI is available on large scale supercomputers, cloud computing systems and it can also be used for interprocess communication on a single compute node -if other shared memory programming models are not  ... 
arXiv:1603.02339v2 fatcat:sff2anv5bfbtfipf4wd5ig75qi

MPI and communication---Adaptive, transparent frequency and voltage scaling of communication phases in MPI programs

Min Yeol Lim, Vincent W. Freeh, David K. Lowenthal
2006 Proceedings of the 2006 ACM/IEEE conference on Supercomputing - SC '06  
This means that the large number of existing MPI programs, as well as new ones being developed, can use our system without modification.  ...  Some microprocessors allow frequency and voltage scaling, which enables a system to reduce CPU performance and power when the CPU is not on the critical path.  ...  MPI Results This section presents our results in three parts. The first part discusses results on several benchmark programs on a power-scalable cluster.  ... 
doi:10.1145/1188455.1188567 dblp:conf/sc/LimFL06 fatcat:2uhpb43dlrdbvnqpgxomnc4bru

Adaptive, Transparent Frequency and Voltage Scaling of Communication Phases in MPI Programs

Min Lim, Vincent Freeh, David Lowenthal
2006 ACM/IEEE SC 2006 Conference (SC'06)  
This means that the large number of existing MPI programs, as well as new ones being developed, can use our system without modification.  ...  Some microprocessors allow frequency and voltage scaling, which enables a system to reduce CPU performance and power when the CPU is not on the critical path.  ...  MPI Results This section presents our results in three parts. The first part discusses results on several benchmark programs on a power-scalable cluster.  ... 
doi:10.1109/sc.2006.11 fatcat:33r5qcfsr5dxtpssynbihrqshy

MegaProto: 1 TFlops/10kW Rack Is Feasible Even with Only Commodity Technology

H. Nakashima, H. Nakamura, M. Sato, T. Boku, S. Matsuoka, D. Takahashi, Y. Hotta
2005 ACM/IEEE SC 2005 Conference (SC'05)  
Performance studies show that even the first version significantly outperforms a conventional high-end 1 U server comprised of dual power-hungry processors in a majority of NPB programs.  ...  In our research project "Mega-Scale Computing Based on Low-Power Technology and Workload Modeling", we claim that a million-scale parallel system could be built with densely mounted low-power commodity  ...  This research work is supported by Japan Science and Technology Agency as a CREST research program entitled "Mega-Scale Computing Based on Low-Power Technology and Workload Modeling."  ... 
doi:10.1109/sc.2005.45 dblp:conf/sc/NakashimaNSBMTH05 fatcat:2nbhh6ewcrgpbi3gzngnvt7v6a

MapReduce for Data Intensive Scientific Analyses

Jaliya Ekanayake, Shrideep Pallickara, Geoffrey Fox
2008 2008 IEEE Fourth International Conference on eScience  
First, we present our experience in applying the MapReduce technique for two scientific data analyses: (i) High Energy Physics data analyses; (ii) Kmeans clustering.  ...  Efficient parallel/concurrent algorithms and frameworks are the key to meeting the scalability and performance requirements entailed in such scientific data analyses.  ...  T 1 is measured using programs, implemented in Java for Hadoop and CGL-MapReduce and C++ for the MPI version, which are run on a single node of the same cluster.  ... 
doi:10.1109/escience.2008.59 dblp:conf/eScience/EkanayakePF08 fatcat:pgrhvxhmrzh7npwhdjvagq27g4

Experiments with Parallelizing Tribology Simulations

V. Chaudhary, W. L. Hase, H. Jiang, L. Sun, D. Thaker
2004 Journal of Supercomputing  
This paper compares OpenMP, MPI, and Strings for parallelizing a complicated tribology problem.  ...  Different parallelization methods vary in their system requirements, programming styles, efficiency of exploring parallelism, and the application characteristics they can handle.  ...  The application was run sequentially, using OpenMP (on the large SMP), using MPI (the MPICH implementation was used) on the cluster of SMPs, and using Strings on the same cluster.  ... 
doi:10.1023/b:supe.0000022103.01620.f3 fatcat:7ns7sjh32zbo5hd4z7aa7zil3m

On the energy footprint of I/O management in Exascale HPC systems

Matthieu Dorier, Orcun Yildiz, Shadi Ibrahim, Anne-Cécile Orgerie, Gabriel Antoniu
2016 Future generations computer systems  
The advent of unprecedentedly scalable yet energy hungry Exascale supercomputers poses a major challenge in sustaining a high performanceper-watt ratio.  ...  Our proposed model gives hints to pre-select the most energy-efficient I/O approach for a particular simulation on a particular HPC system and therefore provides a step towards energy-efficient HPC simulations  ...  The biggest contribution of their work is to be able to apply switch energy gears dynamically by observing the pressure on the memory and the location of MPI calls in the program to obtain better energy  ... 
doi:10.1016/j.future.2016.03.002 fatcat:qgzcoxpja5hkffbuxwzadr3sxa

Optimization of hybrid parallel application execution in heterogeneous high performance computing systems considering execution time and power consumption [article]

Paweł Rościszewski
2018 arXiv   pre-print
Utilizing full power of such systems requires programming parallel applications that are hybrid in two meanings: they can utilize parallelism on multiple levels at the same time and combine together programming  ...  One of the main obstacles to achieving this goal is power consumption of the computing systems that exceeds the energy supply limits.  ...  Numerical Aerodynamic Simulation benchmark [120] on a power-scalable cluster using MPI.  ... 
arXiv:1809.07611v1 fatcat:f2vl3kmgznckroj6h3uwt2zwf4
« Previous Showing results 1 — 15 out of 297 results