Performance Prediction and Calibration for a Class of Multiprocessors
IEEE Transactions on Computers
Parallel applications performance methodology
Instrumentation for future parallel computing systems
A method for performance prediction of parallel programs
CONPAR 90 Proceedings of the joint international conference on Vector and parallel processing
A static performance estimator in the Fortran D programming system
Languages, compilers and run-time environments for distributed memory machines
ICS '92 Proceedings of the 6th international conference on Supercomputing
Analytical performance prediction on multicomputers
Proceedings of the 1993 ACM/IEEE conference on Supercomputing
Interpretive performance prediction for high performance parallel computing
Interpretive performance prediction for high performance parallel computing
Design of a performance technology infrastructure to support the construction of responsive software
Proceedings of the 2nd international workshop on Software and performance
Adaptive Runtime Managementof SAMR Applications
HiPC '02 Proceedings of the 9th International Conference on High Performance Computing
Performance Prediction for Complex Parallel Applications
IPPS '97 Proceedings of the 11th International Symposium on Parallel Processing
| Hi-index | 0.00 |
The authors present an interpretive approach for accurate and cost-effective performance prediction in a high-performance computing environment and describe the design of a compile-time High-Performance Fortran (HPF)/Fortran 90D performance prediction framework based on this approach. The authors have implemented this framework as a part of the HPF/Fortran 90D application-development environment, which integrates it with an HPF/Fortran 90D compiler and a functional interpreter. The current implementation of the environment framework targets the iPSC/860 hypercube multicomputer system. The authors demonstrate how the framework selects appropriate HPF/Fortran 90D compiler directives, debugs application performance, and experiments with runtime and system parameters. First, they explain the interpretive performance-prediction approach, describing the system abstraction module, the application abstraction module, the interpretation engine, and the output module. Then, after a brief description of HPF/Fortran 90D, they discuss the framework's design. The early phases of this design include compilation and interpretation. These phases are followed by the abstraction and interpretation of HPF/Fortran 90D parallel constructs and the abstraction of the iPSC/860 hypercube multicomputer system. Finally, the authors address the steps they've taken to evaluate the framework's accuracy, apply the framework to a high-performance computing environment, and validate its convenience and cost-effectiveness for experimentation involving application development.