HIGH-PERFORMANCE COMPUTING ON HETEROGENEOUS SYSTEMS
Quantum Chemistry Group, Department of Chemistry,
A. Mickiewicz University, Grunwaldzka 6, 60-780 Poznań, Poland
DOI: 10.12921/cmst.1999.05.01.07-19
OAI: oai:lib.psnc.pl:496
Abstract:
A model two-processor heterogeneous computer consisting of one scalar and one vector processor is analyzed in terms of its performance. It is demonstrated that on mixed-type (scalar-vector) applications it is much more effective than a homogeneous environment. Various models of the job distribution are introduced. A working implementation in the field of quantum chemistry is presented.
References:
[1] I. Foster, Designing and Building Parallel Programs (online) (1995). Available at
http ://www. mcs. anl. go v/dbpp/.
[2] C. R. Mechoso, J. D. Farrara, J. A. Spahr, Achieving superlinear speedup on a heterogeneous, distributed system, IEEE Parallel & Distributed Technology 2, 57 (1994).
[3] X. Zhang and Y. Yan, Modeling and characterizing parallel computing performance on
heterogeneous networks of workstations, Proceedings of the Seventh IEEE Symposium on Parallel and Distributed Processing 25 (1995).
[4] W. Cencek and J. Rychlewski, Many-electron explicitly correlated Gaussian functions. I. General theory and test results, J. Chem. Phys. 102, 2533 (1995).
[5] W. Cencek, J. Komasa, J. Rychlewski, Benchmark calculations for two-electron systems using explicitly correlated Gaussian functions, Chem. Phys. Lett. 246, 417 (1995).
A model two-processor heterogeneous computer consisting of one scalar and one vector processor is analyzed in terms of its performance. It is demonstrated that on mixed-type (scalar-vector) applications it is much more effective than a homogeneous environment. Various models of the job distribution are introduced. A working implementation in the field of quantum chemistry is presented.
[1] I. Foster, Designing and Building Parallel Programs (online) (1995). Available at
http ://www. mcs. anl. go v/dbpp/.
[2] C. R. Mechoso, J. D. Farrara, J. A. Spahr, Achieving superlinear speedup on a heterogeneous, distributed system, IEEE Parallel & Distributed Technology 2, 57 (1994).
[3] X. Zhang and Y. Yan, Modeling and characterizing parallel computing performance on
heterogeneous networks of workstations, Proceedings of the Seventh IEEE Symposium on Parallel and Distributed Processing 25 (1995).
[4] W. Cencek and J. Rychlewski, Many-electron explicitly correlated Gaussian functions. I. General theory and test results, J. Chem. Phys. 102, 2533 (1995).
[5] W. Cencek, J. Komasa, J. Rychlewski, Benchmark calculations for two-electron systems using explicitly correlated Gaussian functions, Chem. Phys. Lett. 246, 417 (1995).
