Parallel Processing a Three-Dimensional Free-Lagrange Code: a Case History

Abstract

A three-dimensional, time-dependent free-Lagrange hy drodynamics code has been multitasked and autotasked on a CRAY X-MP/416. The multitasking was done by using the Los Alamos Multitasking Control Library, which is a superset of the CRAY multitasking library. Autotasking is done by using constructs which are only comment cards if the source code is not run through a preprocessor. The three-dimensional algorithm has pre sented a number of problems that simpler algorithms, such as those for one-dimensional hydrodynamics, did not exhibit. Problems in converting the serial code, orig inally written for a CRAY-1, to a multitasking code are discussed. Autotasking of a rewritten version of the code is discussed. Timing results for subroutines and hot spots in the serial code are presented and suggestions for additional tools and debugging aids are given. Theo retical speedup results obtained from Amdahl's law and actual speedup results obtained on a dedicated machine are presented. Suggestions for designing large parallel codes are given.

Get full access to this article

View all access options for this article.

References

Amdahl, G.M. 1967. Validity of the single processor approach to achieving large scale computing capabilities. In AFIPS Conf. Proc. Reston, Va.: AFIPS Press, 30:483-485.

Computer Documentation Group. 1981. TALLY, CTSS-131. Los Alamos National Laboratory.

Noh, W.F. 1983. Artificial viscosity (Q) and artificial heat flux (H) errors for spherically divergent shocks. UCRL-89623. Lawrence Livermore National Laboratory.

Spangenberg, W.H. , and Bobrowicz, F.W. 1986. The memory management system for the Los Alamos CRAY X-MP environment. LA-UR-86-1254, Los Alamos National Laboratory .

Trease, H.E. 1985. Three-dimensional free Lagrangian hydrodynamics . Lecture Notes in Phys. 238:145-157.

Tscharnuter, W.M. , and Winkler, K.H. 1978. A method for computing self-gravitation gas flow with radiation. MPI/PAE-Astro 163.

Von Neumann, J. , and Richtmyer, R.D. 1957. A method for the numerical coding of hydrodynamic shocks . J. Appl. Phys. 21:232.

Ware, W.H. 1972. The ultimate computer. IEEE Spectrum 9(3):84-91.

Williams, E. , and Bobrowicz, F. 1985. Speedup predictions for large scientific parallel programs on CRAY X-MP-like architectures. In Proc. of the 1985 internat. conf. on parallel processing. Washington, D.C.: IEEE Computer Society Press.