skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: From MPI to OpenSHMEM: Porting LAMMPS

 [1];  [1];  [2];  [1];  [2]
  1. University of Tennessee, Knoxville (UTK)
  2. ORNL
Publication Date:
Research Org.:
Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
Sponsoring Org.:
Work for Others (WFO)
OSTI Identifier:
DOE Contract Number:
Resource Type:
Resource Relation:
Conference: OpenSHMEM 2015, Annapolis, MD, USA, 20150804, 20150806
Country of Publication:
United States

Citation Formats

Tang, Chunyan, Herault, Thomas, Bouteiller, Aurelien, Bosilca, George, and Gorentla Venkata, Manjunath. From MPI to OpenSHMEM: Porting LAMMPS. United States: N. p., 2015. Web.
Tang, Chunyan, Herault, Thomas, Bouteiller, Aurelien, Bosilca, George, & Gorentla Venkata, Manjunath. From MPI to OpenSHMEM: Porting LAMMPS. United States.
Tang, Chunyan, Herault, Thomas, Bouteiller, Aurelien, Bosilca, George, and Gorentla Venkata, Manjunath. 2015. "From MPI to OpenSHMEM: Porting LAMMPS". United States. doi:.
title = {From MPI to OpenSHMEM: Porting LAMMPS},
author = {Tang, Chunyan and Herault, Thomas and Bouteiller, Aurelien and Bosilca, George and Gorentla Venkata, Manjunath},
abstractNote = {},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = 2015,
month = 1

Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:
  • LAMMPS is a classical molecular dynamics code, and an acronym for Large-scale Atomic/Molecular Massively Parallel Simulator. LAMMPS has potentials for soft materials (biomolecules, polymers) and solid-state materials (metals, semiconductors) and coarse-grained or mesoscopic systems. It can be used to model atoms or, more generically, as a parallel particle simulator at the atomic, meso, or continuum scale. LAMMPS runs on single processors or in parallel using message-passing techniques and a spatial-decomposition of the simulation domain. The code is designed to be easy to modify or extend with new functionality.
  • Abstract not provided.
  • The OpenSHMEM reference implementation was developed towards the goal of developing an open source and high-performing Open- SHMEM implementation. To achieve portability and performance across various networks, the OpenSHMEM reference implementation uses GAS- Net and UCCS for network operations. Recently, new network layers have emerged with the promise of providing high-performance, scalabil- ity, and portability for HPC applications. In this paper, we implement the OpenSHMEM reference implementation to use the UCX framework for network operations. Then, we evaluate its performance and scalabil- ity on Cray XK systems to understand UCX s suitability for developing the OpenSHMEM programming model. Further, wemore » develop a bench- mark called SHOMS for evaluating the OpenSHMEM implementation. Our experimental results show that OpenSHMEM-UCX outperforms the vendor supplied OpenSHMEM implementation in most cases on the Cray XK system by up to 40% with respect to message rate and up to 70% for the execution of application kernels.« less
  • High-speed cine photography has been applied to an optically-accessed version of a modern 4-valve, pent-roof, SI engine in order to observe the behavior of fuel under cold starting and warm-up conditions. A wall film was formed on the combustion chamber surface in the vicinity of the valves when fuel was supplied by pintle and disk-type injectors. This wall film survived combustion but was drawn into the exhaust system when the exhaust valves opened. The use of an air-blast atomizer avoided the formation of this wall film, but some wetting of the piston was apparent, as it was with the othermore » types of injector.« less
  • CTH is a family of codes developed at Sandia National Laboratories for use in modeling complex multi-dimensional, multi-material problems that are characterized by large deformations and/or strong shocks. A two-step, second-order accurate Eulerian solution algorithm is used to solve the mass, momentum, and energy conservation equations. CTH has historically been run on systems where the data are directly accessible to the cpu, such as workstations and vector supercomputers. Multiple cpus can be used if all data are accessible to all cpus. This is accomplished by placing compiler directives or subroutine calls within the source code. The CTH team has implementedmore » this scheme for Cray shared memory machines under the Unicos operating system. This technique is effective, but difficult to port to other (similar) shared memory architectures because each vendor has a different format of directives or subroutine calls. A different model of high performance computing is one where many (> 1,000) cpus work on a portion of the entire problem and communicate by passing messages that contain boundary data. Most, if not all, codes that run effectively on parallel hardware were written with a parallel computing paradigm in mind. Modifying an existing code written for serial nodes poses a significantly different set of challenges that will be discussed. CTH, a legacy FORTRAN code, has been modified to allow for solutions on distributed memory parallel computers such as the IBM SP2, the Intel Paragon, Cray T3D, or a network of workstations. The message passing version of CTH will be discussed and example calculations will be presented along with performance data. Current timing studies indicate that CTH is 2--3 times faster than equivalent C++ code written specifically for parallel hardware. CTH on the Intel Paragon exhibits linear speed up with problems that are scaled (constant problem size per node) for the number of parallel nodes.« less