Large-scale scientific application programs in chemistry and physics on an experimental parallel computer system

Corongiu, G.; Detrich, John

doi:10.1147/rd.294.0422

Cited by 19 publications

(7 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Multiple FPS attached processors can be used in a single job in parallel. [26][27][28] In the parallel computations the actual evaluation of forces is parceled out to several independent APs by passing the information of the coordinates and orientations of the particles to each AP. Each AP is then assigned only a partial number of force evaluations to perform.…”

Section: Parallel Computationsmentioning

confidence: 99%

Hydration structure and dynamics of B‐ and Z‐DNA in the presence of counterions via molecular dynamics simulations

Clementi

1987

Biopolymers

View full text Add to dashboard Cite

SynopsisFollowing our previous attempts at understanding the structural and dynamical properties of water and counterions hydrating nucleic acids, we have performed molecular dynamics simulations for B-and Z-DNA. In these simulations, the nucleic acids were held rigid. In the case of B-DNA, one turn of B-DNA double helix was considered in the presence of 1500 water molecules and 20 counterions (K+). The simulations were performed for 4.0 ps after equilibrating the system. For Z-DNA, we considered one turn of the double helix in the presence of 1851 water molecules and 24 counterions (K+). The simulations were carried out for 3.5 ps after equilibration. The average temperature of these simulations was -360 K for Z-DNA and -345 K for B-DNA. In these simulations the hydrogen atoms were explicitly taken into account. For both simulations, a fifth-order predictor-corrector was used for solving the translational equations of motion. The rotational motion of the water molecules was represented in terms of quaternion algebra and the rotational equations of motion were solved with a second-order quaternion method using a sixth-order predictor-corrector method. A time step of 0.5 . s was used in these simulations. The structural and the dynamical properties of water solvating the counterions, and the phosphate groups of the DNA, were computed to understand the hydration structure. Diffusion coefficients and velocity correlation functions were calculated for both ions and the water molecules. The velocity correlation functions for the ions exhibit a caged behavior. The dipole correlation functions for the water molecules indicate that the water molecules close to the helix retain the memory of their initial orientations for longer periods of time than those away from the helix. During the time period of our simulation (3-4 ps) the ion probability distributions show a well-defined pattern and suggest limited mobility for the ions, being clase to the helix.

show abstract

Section: Parallel Computationsmentioning

confidence: 99%

Hydration structure and dynamics of B‐ and Z‐DNA in the presence of counterions via molecular dynamics simulations

Clementi

1987

Biopolymers

View full text Add to dashboard Cite

show abstract

“…We understood beforehand exactly the nature of the parallelism inherent in the calculations and the large grain parallelism involved. A very simple but operational master-slave virtual machine-(VM) based operating system called VMFACS [13] was in use within 3 months. The possibility of using the two or four engines of the IBM-3081 and IBM-3084 in parallel was also immediately considered; this effort resulted in a very simple but operational MVS-based operating system called PARADIGM [ l l ] which, in late 1983 and early 1984, was used for Monte Carlo as well as for quantum mechanical computations.…”

Section: Architecturementioning

confidence: 99%

Supercomputing and supercomputers for science and engineering in general and for chemistry and biosciences in particular

Clementi

Chin

Corongiu

et al. 1989

Int J of Quantum Chemistry

Self Cite

View full text Add to dashboard Cite

We start by pointing out relationships between production of information, global simulation, and supercomputing, thus placing our research activities in today's society context. Then we detail the evolution in hardware and software for lCAP, our experimental supercomputer, which we claim to be especially well suited for supercomputing in science and engineering. A preliminary discussion of 1CAP/3090 (our latest experimental effort) is included. Many examples from different disciplines are provided to verify our assertions. We "prove" our point by presenting an example of global supercomputing. Starting with 3 nuclei and 10 electrons, building up to a single water molecule, then to a few hundred, we learn, for example, about Raman, infrared, and neutron scattering; we then move up to a few hundred thousand molecules to analyze particle flow and obstructions; finally we experiment, but only preliminarily, with a few million particles to learn more on nonequilibrium dynamics as in the Rayleigh-Benard systems. In this way, quantum mechanics is overlapped with statistical mechanics and expanded into microdynamics. The entire paper is finally reanalyzed from a different perspective, presenting rather systematically, even if most briefly, our ideas on "modern" computational chemistry, where quantum mechanics is as much needed as fluid dynamics and graphics. In this section the main computational techniques are analyzed in terms of computer programs and their associated flow diagrams to solve the basic equations using parallel supercomputers.-

show abstract

“…The system currently in use in our laboratory consists of an IBM host computer (an IBM 4341 at present) and a number of attached processors (six FPS-164's at present). With a theoretical maximum of 12 MFLOPS, the FPS-164 is a rather powerful processor for scientific calculations, but it still falls somewhat short of the computing power required for the calculations reported here, as well as other large-scale calculations being carried out in our laboratory [4,5]. Our answer to this difficulty is an experimental system which allows all of the attached processors (APs) to cooperate on a single computational task, provided that the task in question can be divided into subtasks for the attached processors to perform simultaneously.…”

Section: Program Migration Considerationsmentioning

confidence: 99%

“…The practical implementation of this system requires that the main program (which runs on the host) be able to closely control the subtasks running on the APs, periodically transferring data to them, starting them, and subsequently gathering up each set of subtask output. For such system design considerations, we refer the reader elsewhere [4]. Here we take the system as given, and present what we have learned about migration of our Metropolis-Monte Carlo liquid simulation programs to utilize this system effectively.…”

Section: Program Migration Considerationsmentioning

confidence: 99%

See 1 more Smart Citation

Monte Carlo liquid water simulations with four-body interactions included

Corongiu¹,

Clementi²

1984

Int. J. Quantum Chem.

View full text Add to dashboard Cite

A number of our Metropolis–Monte Carlo liquid simulation programs have been successfully modified to run on a system consisting of an IBM host computer and several array processors (FPS‐164) configured to allow execution of a single computation on multiple attached processors. These include programs which incorporate three‐ and four‐body molecular interaction terms, and therefore require a high‐performance computing system for practical execution. The strategies used to modify our Fortran programs are discussed. Preliminary results from practical calculations using these programs are presented to demonstrate the performance gains achieved by our configuration for large‐scale scientific computations; in particular, the techniques developed for this Monte Carlo study and for quantum‐chemical computations have already been successfully adopted in converting molecular dynamics simulations to this configuration.

show abstract

Large-scale scientific application programs in chemistry and physics on an experimental parallel computer system

Cited by 19 publications

References 0 publications

Hydration structure and dynamics of B‐ and Z‐DNA in the presence of counterions via molecular dynamics simulations

Hydration structure and dynamics of B‐ and Z‐DNA in the presence of counterions via molecular dynamics simulations

Supercomputing and supercomputers for science and engineering in general and for chemistry and biosciences in particular

Monte Carlo liquid water simulations with four-body interactions included

Contact Info

Product

Resources

About