# Pthreads vs MPI Parallel Performance of Angular-Domain Decomposed S

## Abstract

Two programming models for parallelizing the Angular Domain Decomposition (ADD) of the discrete ordinates (S{sub n}) approximation of the neutron transport equation are examined. These are the shared memory model based on the POSIX threads (Pthreads) standard, and the message passing model based on the Message Passing Interface (MPI) standard. These standard libraries are available on most multiprocessor platforms thus making the resulting parallel codes widely portable. The question is: on a fixed platform, and for a particular code solving a given test problem, which of the two programming models delivers better parallel performance? Such comparison is possible on Symmetric Multi-Processors (SMP) architectures in which several CPUs physically share a common memory, and in addition are capable of emulating message passing functionality. Implementation of the two-dimensional,(S{sub n}), Arbitrarily High Order Transport (AHOT) code for solving neutron transport problems using these two parallelization models is described. Measured parallel performance of each model on the COMPAQ AlphaServer 8400 and the SGI Origin 2000 platforms is described, and comparison of the observed speedup for the two programming models is reported. For the case presented in this paper it appears that the MPI implementation scales better than the Pthreads implementation on both platforms.

- Authors:

- Publication Date:

- Research Org.:
- Oak Ridge National Lab., TN (US)

- Sponsoring Org.:
- USDOE Office of Science (US)

- OSTI Identifier:
- 754815

- Report Number(s):
- ORNL/P00-106228

TRN: US0005062

- DOE Contract Number:
- AC05-00OR22725

- Resource Type:
- Conference

- Resource Relation:
- Conference: Physor 2000 ANS International Topical Meeting, Pittsburgh, PA (US), 05/07/2000--05/11/2000; Other Information: PBD: 7 May 2000

- Country of Publication:
- United States

- Language:
- English

- Subject:
- 73 NUCLEAR PHYSICS AND RADIATION PHYSICS; 99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; ARRAY PROCESSORS; DISCRETE ORDINATE METHOD; IMPLEMENTATION; NEUTRON TRANSPORT; PARALLEL PROCESSING; NEUTRON TRANSPORT THEORY

### Citation Formats

```
Azmy, Y.Y., and Barnett, D.A.
```*Pthreads vs MPI Parallel Performance of Angular-Domain Decomposed S*. United States: N. p., 2000.
Web.

```
Azmy, Y.Y., & Barnett, D.A.
```*Pthreads vs MPI Parallel Performance of Angular-Domain Decomposed S*. United States.

```
Azmy, Y.Y., and Barnett, D.A. Sun .
"Pthreads vs MPI Parallel Performance of Angular-Domain Decomposed S". United States. https://www.osti.gov/servlets/purl/754815.
```

```
@article{osti_754815,
```

title = {Pthreads vs MPI Parallel Performance of Angular-Domain Decomposed S},

author = {Azmy, Y.Y. and Barnett, D.A.},

abstractNote = {Two programming models for parallelizing the Angular Domain Decomposition (ADD) of the discrete ordinates (S{sub n}) approximation of the neutron transport equation are examined. These are the shared memory model based on the POSIX threads (Pthreads) standard, and the message passing model based on the Message Passing Interface (MPI) standard. These standard libraries are available on most multiprocessor platforms thus making the resulting parallel codes widely portable. The question is: on a fixed platform, and for a particular code solving a given test problem, which of the two programming models delivers better parallel performance? Such comparison is possible on Symmetric Multi-Processors (SMP) architectures in which several CPUs physically share a common memory, and in addition are capable of emulating message passing functionality. Implementation of the two-dimensional,(S{sub n}), Arbitrarily High Order Transport (AHOT) code for solving neutron transport problems using these two parallelization models is described. Measured parallel performance of each model on the COMPAQ AlphaServer 8400 and the SGI Origin 2000 platforms is described, and comparison of the observed speedup for the two programming models is reported. For the case presented in this paper it appears that the MPI implementation scales better than the Pthreads implementation on both platforms.},

doi = {},

journal = {},

number = ,

volume = ,

place = {United States},

year = {2000},

month = {5}

}