The Experts below are selected from a list of 44394 Experts worldwide ranked by ideXlab platform
Michael A. Frumkin - One of the best experts on this subject based on the ideXlab platform.
-
Automatic Recognition of Performance Idioms in Scientific Applications
2011 IEEE International Parallel & Distributed Processing Symposium, 2011Co-Authors: Jiahua He, Allan E. Snavely, Rob F. Van Der Wijngaart, Michael A. FrumkinAbstract:Basic data flow patterns that we call performance Idioms, such as stream, transpose, reduction, random access and stencil, are common in scientific numerical applications. We hypothesize that a small number of Idioms can cover most programming constructs that dominate the execution time of scientific codes and can be used to approximate the application performance. To check these hypotheses, we proposed an automatic Idioms recognition method and implemented the method, based on the open source compiler Open64. With the NAS Parallel Benchmark (NPB) as a case study, the prototype system is about 90% accurate compared with idiom classification by a human expert. Our results showed that the above five Idioms suffice to cover 100% of the six NPB codes (MG, CG, FT, BT, SP and LU). We also compared the performance of our idiom benchmarks with their corresponding instances in the NPB codes on two different platforms with different methods. The approximation accuracy is up to 96.6%. The contribution is to show that a small set of Idioms can cover more complex codes, that Idioms can be recognized automatically, and that suitably defined Idioms may approximate application performance.
-
automatic recognition of performance Idioms in scientific applications
International Parallel and Distributed Processing Symposium, 2011Co-Authors: Jiahua He, Allan E. Snavely, Rob F. Van Der Wijngaart, Michael A. FrumkinAbstract:Basic data flow patterns that we call \textbf{performance Idioms}, such as stream, transpose, reduction, random access and stencil, are common in scientific numerical applications. We hypothesize that a small number of Idioms can cover most programming constructs that dominate the execution time of scientific codes and can be used to approximate the application performance. To check these hypotheses, we proposed an automatic Idioms recognition method and implemented the method, based on the open source compiler Open64. With the NAS Parallel Benchmark (NPB) as a case study, the prototype system is about $90%$ accurate compared with idiom classification by a human expert. Our results showed that the above five Idioms suffice to cover $100%$ of the six NPB codes (MG, CG, FT, BT, SP and LU). We also compared the performance of our idiom benchmarks with their corresponding instances in the NPB codes on two different platforms with different methods. The approximation accuracy is up to $96.6%$. The contribution is to show that a small set of Idioms can cover more complex codes, that Idioms can be recognized automatically, and that suitably defined Idioms may approximate application performance.
Jiahua He - One of the best experts on this subject based on the ideXlab platform.
-
Automatic Recognition of Performance Idioms in Scientific Applications
2011 IEEE International Parallel & Distributed Processing Symposium, 2011Co-Authors: Jiahua He, Allan E. Snavely, Rob F. Van Der Wijngaart, Michael A. FrumkinAbstract:Basic data flow patterns that we call performance Idioms, such as stream, transpose, reduction, random access and stencil, are common in scientific numerical applications. We hypothesize that a small number of Idioms can cover most programming constructs that dominate the execution time of scientific codes and can be used to approximate the application performance. To check these hypotheses, we proposed an automatic Idioms recognition method and implemented the method, based on the open source compiler Open64. With the NAS Parallel Benchmark (NPB) as a case study, the prototype system is about 90% accurate compared with idiom classification by a human expert. Our results showed that the above five Idioms suffice to cover 100% of the six NPB codes (MG, CG, FT, BT, SP and LU). We also compared the performance of our idiom benchmarks with their corresponding instances in the NPB codes on two different platforms with different methods. The approximation accuracy is up to 96.6%. The contribution is to show that a small set of Idioms can cover more complex codes, that Idioms can be recognized automatically, and that suitably defined Idioms may approximate application performance.
-
automatic recognition of performance Idioms in scientific applications
International Parallel and Distributed Processing Symposium, 2011Co-Authors: Jiahua He, Allan E. Snavely, Rob F. Van Der Wijngaart, Michael A. FrumkinAbstract:Basic data flow patterns that we call \textbf{performance Idioms}, such as stream, transpose, reduction, random access and stencil, are common in scientific numerical applications. We hypothesize that a small number of Idioms can cover most programming constructs that dominate the execution time of scientific codes and can be used to approximate the application performance. To check these hypotheses, we proposed an automatic Idioms recognition method and implemented the method, based on the open source compiler Open64. With the NAS Parallel Benchmark (NPB) as a case study, the prototype system is about $90%$ accurate compared with idiom classification by a human expert. Our results showed that the above five Idioms suffice to cover $100%$ of the six NPB codes (MG, CG, FT, BT, SP and LU). We also compared the performance of our idiom benchmarks with their corresponding instances in the NPB codes on two different platforms with different methods. The approximation accuracy is up to $96.6%$. The contribution is to show that a small set of Idioms can cover more complex codes, that Idioms can be recognized automatically, and that suitably defined Idioms may approximate application performance.
Allan E. Snavely - One of the best experts on this subject based on the ideXlab platform.
-
Automatic Recognition of Performance Idioms in Scientific Applications
2011 IEEE International Parallel & Distributed Processing Symposium, 2011Co-Authors: Jiahua He, Allan E. Snavely, Rob F. Van Der Wijngaart, Michael A. FrumkinAbstract:Basic data flow patterns that we call performance Idioms, such as stream, transpose, reduction, random access and stencil, are common in scientific numerical applications. We hypothesize that a small number of Idioms can cover most programming constructs that dominate the execution time of scientific codes and can be used to approximate the application performance. To check these hypotheses, we proposed an automatic Idioms recognition method and implemented the method, based on the open source compiler Open64. With the NAS Parallel Benchmark (NPB) as a case study, the prototype system is about 90% accurate compared with idiom classification by a human expert. Our results showed that the above five Idioms suffice to cover 100% of the six NPB codes (MG, CG, FT, BT, SP and LU). We also compared the performance of our idiom benchmarks with their corresponding instances in the NPB codes on two different platforms with different methods. The approximation accuracy is up to 96.6%. The contribution is to show that a small set of Idioms can cover more complex codes, that Idioms can be recognized automatically, and that suitably defined Idioms may approximate application performance.
-
automatic recognition of performance Idioms in scientific applications
International Parallel and Distributed Processing Symposium, 2011Co-Authors: Jiahua He, Allan E. Snavely, Rob F. Van Der Wijngaart, Michael A. FrumkinAbstract:Basic data flow patterns that we call \textbf{performance Idioms}, such as stream, transpose, reduction, random access and stencil, are common in scientific numerical applications. We hypothesize that a small number of Idioms can cover most programming constructs that dominate the execution time of scientific codes and can be used to approximate the application performance. To check these hypotheses, we proposed an automatic Idioms recognition method and implemented the method, based on the open source compiler Open64. With the NAS Parallel Benchmark (NPB) as a case study, the prototype system is about $90%$ accurate compared with idiom classification by a human expert. Our results showed that the above five Idioms suffice to cover $100%$ of the six NPB codes (MG, CG, FT, BT, SP and LU). We also compared the performance of our idiom benchmarks with their corresponding instances in the NPB codes on two different platforms with different methods. The approximation accuracy is up to $96.6%$. The contribution is to show that a small set of Idioms can cover more complex codes, that Idioms can be recognized automatically, and that suitably defined Idioms may approximate application performance.
Rob F. Van Der Wijngaart - One of the best experts on this subject based on the ideXlab platform.
-
Automatic Recognition of Performance Idioms in Scientific Applications
2011 IEEE International Parallel & Distributed Processing Symposium, 2011Co-Authors: Jiahua He, Allan E. Snavely, Rob F. Van Der Wijngaart, Michael A. FrumkinAbstract:Basic data flow patterns that we call performance Idioms, such as stream, transpose, reduction, random access and stencil, are common in scientific numerical applications. We hypothesize that a small number of Idioms can cover most programming constructs that dominate the execution time of scientific codes and can be used to approximate the application performance. To check these hypotheses, we proposed an automatic Idioms recognition method and implemented the method, based on the open source compiler Open64. With the NAS Parallel Benchmark (NPB) as a case study, the prototype system is about 90% accurate compared with idiom classification by a human expert. Our results showed that the above five Idioms suffice to cover 100% of the six NPB codes (MG, CG, FT, BT, SP and LU). We also compared the performance of our idiom benchmarks with their corresponding instances in the NPB codes on two different platforms with different methods. The approximation accuracy is up to 96.6%. The contribution is to show that a small set of Idioms can cover more complex codes, that Idioms can be recognized automatically, and that suitably defined Idioms may approximate application performance.
-
automatic recognition of performance Idioms in scientific applications
International Parallel and Distributed Processing Symposium, 2011Co-Authors: Jiahua He, Allan E. Snavely, Rob F. Van Der Wijngaart, Michael A. FrumkinAbstract:Basic data flow patterns that we call \textbf{performance Idioms}, such as stream, transpose, reduction, random access and stencil, are common in scientific numerical applications. We hypothesize that a small number of Idioms can cover most programming constructs that dominate the execution time of scientific codes and can be used to approximate the application performance. To check these hypotheses, we proposed an automatic Idioms recognition method and implemented the method, based on the open source compiler Open64. With the NAS Parallel Benchmark (NPB) as a case study, the prototype system is about $90%$ accurate compared with idiom classification by a human expert. Our results showed that the above five Idioms suffice to cover $100%$ of the six NPB codes (MG, CG, FT, BT, SP and LU). We also compared the performance of our idiom benchmarks with their corresponding instances in the NPB codes on two different platforms with different methods. The approximation accuracy is up to $96.6%$. The contribution is to show that a small set of Idioms can cover more complex codes, that Idioms can be recognized automatically, and that suitably defined Idioms may approximate application performance.
Raymond W. Gibbs - One of the best experts on this subject based on the ideXlab platform.
-
Metaphor in Idiom Comprehension
Journal of Memory and Language, 1997Co-Authors: Raymond W. Gibbs, Josephine M. Bogdanovich, Jeffrey R. Sykes, Dale J. BarrAbstract:Psycholinguistic research has shown that people's tacit knowledge of conceptual metaphors, such as ANGER IS HEATED FLUID IN A CONTAINER, partly motivates how they make sense of idiomatic phrases likeblow your stackandflip your lid.But do people quickly access conceptual metaphors each time an idiom is encountered in discourse? The present studies used a priming method to examine the role of conceptual metaphors in immediate idiom comprehension. Experiment 1 showed that people access conceptual metaphors when understanding Idioms, but significantly less so when processing literal paraphrases of Idioms. Experiment 2 demonstrated that people access the appropriate conceptual metaphors, such as ANGER IS HEAT, when processing some Idioms, such asblow your stack,but not when they read Idioms, such asjump down your throat,which have similar figurative meanings that are motivated by different conceptual metaphors (e.g., ANGER IS ANIMAL BEHAVIOR). The findings from these studies provide important evidence on the constraining role that common patterns of metaphoric thought have in figurative language understanding.
-
what do Idioms really mean
Journal of Memory and Language, 1992Co-Authors: Raymond W. GibbsAbstract:The “dead” metaphor view of idiomaticity suggests that Idioms were once metaphoric but have lost their metaphoricity over time and now are equivalent to simple literal phrases such that blow your stack = “to get very angry,” crack the whip = “to exert authority,” and spill the beans = “to reveal a secret.” The purpose of the present studies was to demonstrate that Idioms are not dead metaphors but have more complex meanings that are motivated by conceptual metaphors linking idiom phrases with their figurative interpretations. Six experiments are reported that examine the difference between Idioms and their literal paraphrases. A first study examined people's intuitions about different knowledge domains to illustrate that the meanings of Idioms are consistent with the source-to-target domain mappings of the conceptual metaphors that motivate these phrases' figurative meanings. The data from Experiments 2 and 3 indicate that people view Idioms as having more complex meanings than do their roughly, equivalent literal paraphrases. Experiments 4 through 6 show that Idioms are most appropriate to use and easiest to comprehend when they are encountered in discourse situations that are consistent with the entailments of the conceptual metaphors that motivate these phrases' idiomatic meanings. The findings from these studies suggest that Idioms are not dead metaphors with simple figurative interpretations. Instead, Idioms have complex meanings that are motivated by independently existing conceptual metaphors that are partly constitutive of everyday thought.
-
Conceptual knowledge in the interpretation of Idioms.
Journal of Experimental Psychology: General, 1990Co-Authors: Nandini P. Nayak, Raymond W. GibbsAbstract:: The authors examined how people determine the contextual appropriateness of Idioms. In Experiment 1, Idioms referring to the same temporal stage of a conceptual prototype were judged to be more similar in meaning than Idioms referring to different temporal stages. In Experiment 2, Idioms in a prototypical temporal sequence were more meaningful than Idioms in sentences that violated the temporal sequence. In Experiment 3, Idioms referring to the same stage of a conceptual prototype were differentiable on the basis of conceptual information. The conceptual coherence between Idioms and contexts facilitated the processing speed of Idioms in Experiment 4. Experiment 5 showed that speakers can recover the underlying conceptual metaphors that link an idiom to its figurative meaning. Experiment 6 showed that the metaphoric information reflected in the lexical makeup of Idioms also determined the metaphoric appropriateness of Idioms in certain contexts.
