Performance prediction based on inherent program similarity Export

@inproceedings{citeulike:5209379, abstract = {A key challenge in benchmarking is to predict the performance of an application of interest on a number of platforms in order to determine which platform yields the best performance. This paper proposes an approach for doing this. We measure a number of microarchitecture-independent characteristics from the application of interest, and relate these characteristics to the characteristics of the programs from a previously profiled benchmark suite. Based on the similarity of the application of interest with programs in the benchmark suite, we make a performance prediction of the application of interest. We propose and evaluate three approaches (normalization, principal components analysis and genetic algorithm) to transform the raw data set of microarchitecture-independent characteristics into a benchmark space in which the relative distance is a measure for the relative performance differences. We evaluate our approach using all of the SPEC CPU2000 benchmarks and real hardware performance numbers from the SPEC website. Our framework estimates per-benchmark machine ranks with a 0.89 average and a 0.80 worst case rank correlation coefficient.}, address = {New York, NY, USA}, author = {Hoste, Kenneth and Phansalkar, Aashish and Eeckhout, Lieven and Georges, Andy and John, Lizy K. and De Bosschere, Koen}, booktitle = {PACT '06: Proceedings of the 15th international conference on Parallel architectures and compilation techniques}, citeulike-article-id = {5209379}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1152174}, citeulike-linkout-1 = {http://dx.doi.org/10.1145/1152154.1152174}, doi = {10.1145/1152154.1152174}, isbn = {1-59593-264-X}, keywords = {dynamic\_analysis, intel\_pin, performance\_prediction}, location = {Seattle, Washington, USA}, pages = {114--122}, posted-at = {2009-07-20 18:54:22}, priority = {2}, publisher = {ACM}, title = {Performance prediction based on inherent program similarity}, url = {http://dx.doi.org/10.1145/1152154.1152174}, year = {2006} }

TY - CONF ID - citeulike:5209379 L3 - citeulike-article-id:5209379 N2 - A key challenge in benchmarking is to predict the performance of an application of interest on a number of platforms in order to determine which platform yields the best performance. This paper proposes an approach for doing this. We measure a number of microarchitecture-independent characteristics from the application of interest, and relate these characteristics to the characteristics of the programs from a previously profiled benchmark suite. Based on the similarity of the application of interest with programs in the benchmark suite, we make a performance prediction of the application of interest. We propose and evaluate three approaches (normalization, principal components analysis and genetic algorithm) to transform the raw data set of microarchitecture-independent characteristics into a benchmark space in which the relative distance is a measure for the relative performance differences. We evaluate our approach using all of the SPEC CPU2000 benchmarks and real hardware performance numbers from the SPEC website. Our framework estimates per-benchmark machine ranks with a 0.89 average and a 0.80 worst case rank correlation coefficient. CY - Seattle, Washington, USA SN - 1-59593-264-X AD - New York, NY, USA EP - 122 TI - Performance prediction based on inherent program similarity PB - ACM T2 - PACT '06: Proceedings of the 15th international conference on Parallel architectures and compilation techniques SP - 114 KW - dynamic_analysis KW - intel_pin KW - performance_prediction AU - Hoste, Kenneth AU - Phansalkar, Aashish AU - Eeckhout, Lieven AU - Georges, Andy AU - John, Lizy AU - De Bosschere, Koen PY - 2006/// UR - http://dx.doi.org/10.1145/1152154.1152174 ER -