Skeleton based performance prediction on shared networks

Sodhi, S.; Subhlok, Jaspal

doi:10.1109/ccgrid.2004.1336704

Cited by 13 publications

(10 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We initialize the pointers to the top of each trace (line 1-3). Then, we tackle the function conflicts and merge similar events (line [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]. If the corresponding events can be merged without function conflicts, they are simply joined into the final trace.…”

Section: Trace-merging Algorithmmentioning

confidence: 99%

“…It models the target architecture with a configuration file. Meanwhile, Sodhi and Subhlok [6], [7] propose a framework for automatic generation of performance skeletons. Xu et al [8], [9] present generation of coordinated performance skeletons, similar to dwarf code with logicalization and compression procedures.…”

Section: Related Workmentioning

confidence: 99%

“…1) Trace merging：Several studies, for example, [6], [7], [8], [11], [12], have been conducted on trace merging algorithms but they all have some pitfalls. Sodhi et al [6], [7] put forward the methods to identify and cluster similar events without considering sequence differences and function conflicts.…”

Section: Related Workmentioning

confidence: 99%

“…Although several related studies have been conducted and well-grounded in trace recording and code generation [5], [6], [7], [8], [9], [10], [11], [12], challenges remain in trace merging and repeat compression. In order to accomplish this, first, we need to tackle function conflicts in trace merging with lower time complexity.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

DwarfCode: A Performance Prediction Tool for Parallel Applications

Zhang

Cheng

Subhlok

2016

IEEE Trans. Comput.

View full text Add to dashboard Cite

We present DwarfCode, a performance prediction tool for MPI applications on diverse computing platforms. The goal is to accurately predict the running time of applications for task scheduling and job migration. First, DwarfCode collects the execution traces to record the computing and communication events. Then, it merges the traces from different processes into a single trace. After that, DwarfCode identifies and compresses the repeating patterns in the final trace to shrink the size of the events. Finally, a dwarf code is generated to mimic the original program behavior. This smaller running benchmark is replayed in the target platform to predict the performance of the original application. In order to generate such a benchmark, two major challenges are to reduce the time complexity of trace merging and repeat compression algorithms. We propose an O(mpn) trace merging algorithm to combine the traces generated by separate MPI processes , where m denotes the upper bound of tracing distance, p denotes the number of processes, and n denotes the maximum of event numbers of all the traces. More importantly, we put forward a novel repeat compression algorithm, whose time complexity is O(nlogn). Experimental results show that DwarfCode can accurately predict the running time of MPI applications. The error rate is below 10% for compute and communication intensive applications. This toolkit has been released for free download as a GNU General Public License v3 software.

show abstract

Section: Trace-merging Algorithmmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

DwarfCode: A Performance Prediction Tool for Parallel Applications

Zhang

Cheng

Subhlok

2016

IEEE Trans. Comput.

View full text Add to dashboard Cite

show abstract

“…The fundamental goal is to build and validate a framework for automatic construction of performance skeletons for parallel MPI programs. Earlier work in this project constructed and validated memory and performance skeletons and explored their usage in distributed environments [3,7,4,9,5]. A new procedure for construction of performance skeletons that is being employed now is illustrated in Figure 1 and detailed in [8].…”

Section: Overviewmentioning

confidence: 99%

Automatic construction of coordinated performance skeletons

Subhlok

2008

2008 IEEE International Symposium on Parallel and Distributed Processing

View full text Add to dashboard Cite

OverviewThis research is motivated by the problem of performance prediction in dynamic and unpredictable environments where traditional performance modeling has limited success. Examples include shared execution environments and new software or system environments. The approach is based on the concept of a performance skeleton which is a short running program whose execution time in any scenario reflects the estimated execution time of the application it represents. The fundamental goal is to build and validate a framework for automatic construction of performance skeletons for parallel MPI programs. Earlier work in this project constructed and validated memory and performance skeletons and explored their usage in distributed environments [3,7,4,9,5]. A new procedure for construction of performance skeletons that is being employed now is illustrated in Figure 1 and detailed in [8]. This paper outlines the new contributions of this work which are 1) trace logicalization which is conversion of a suite of execution traces of an SPMD MPI program into a single logical trace, 2) trace compression which involves identification of the loop structure inherent in the execution trace to capture the core execution behavior, and (3) skeleton construction and validation.

show abstract

Synapse: Synthetic application profiler and emulator

Merzky

Turilli

et al. 2018

Journal of Computational Science

View full text Add to dashboard Cite

Motivated by the need to emulate workload execution characteristics on highperformance and distributed heterogeneous resources, we introduce Synapse. Synapse is used as a proxy application (or "representative application") for real workloads, with the advantage that it can be tuned in different ways and dimensions, and also at levels of granularity that are not possible with real applications. Synapse has a platform-independent application profiler, and has the ability to emulate profiled workloads on a variety of resources. Experiments show that the automated profiling performed using Synapse captures an application's characteristics with high fidelity. The emulation of an application using Synapse can reproduce the application's execution behavior in the original runtime environment, and can also reproduce those behaviors on different run-time environments.

show abstract

Skeleton based performance prediction on shared networks

Cited by 13 publications

References 23 publications

DwarfCode: A Performance Prediction Tool for Parallel Applications

DwarfCode: A Performance Prediction Tool for Parallel Applications

Automatic construction of coordinated performance skeletons

Synapse: Synthetic application profiler and emulator

Contact Info

Product

Resources

About