Parallel Processor System

This article reviews some of the progress made in using parallel processor systems to study macromolecules. After an initial introduction to the key concepts required to understand parallelisation, the main part of the article focuses on molecular dynamics. It is shown that simple replicated data methods can be used to carry out molecular dynamics effectively, without the need for major changes from the approach used in scalar codes. Domain decomposition methods are then introduced as a path toward reducing inter-processor communication costs further to produce truly scalable simulation algorithms. Finally, some of the methods available for carrying out parallel Monte Carlo simulations are discussed. 

One reason for the development of parallel processor systems is to increase the performance and to overcome the limits of a sequential single processor system. (Brinch and Hansen, 1973 Dijkstra, 1968 Gargantini, 1982 Giloi, 1981 Hoener and Roehder, 1976, 1977)... 

Ameling W, Hoener S, Roehder W (1977) Interconnection structures for parallel processor system. 

The rapid rise in computer speed over recent years has led to atom-based simulations of liquid crystals becoming an important new area of research. Molecular mechanics and Monte Carlo studies of isolated liquid crystal molecules are now routine. However, care must be taken to model properly the influence of a nematic mean field if information about molecular structure in a mesophase is required. The current state-of-the-art consists of studies of (in the order of) 100 molecules in the bulk, in contact with a surface, or in a bilayer in contact with a solvent. Current simulation times can extend to around 10 ns and are sufficient to observe the growth of mesophases from an isotropic liquid. The results from a number of studies look very promising, and a wealth of structural and dynamic data now exists for bulk phases, monolayers and bilayers. Continued development of force fields for liquid crystals will be particularly important in the next few years, and particular emphasis must be placed on the development of all-atom force fields that are able to reproduce liquid phase densities for small molecules. Without these it will be difficult to obtain accurate phase transition temperatures. It will also be necessary to extend atomistic models to several thousand molecules to remove major system size effects which are present in all current work. This will be greatly facilitated by modern parallel simulation methods that allow molecular dynamics simulations to be carried out in parallel on multi-processor systems [115]. 

An independent but complementary approach has been the development of parallel computing systems, where many processors can be concurrently applied in cooperation on a single calculation. This idea is certainly not new. It has already been the subject of numerous research projects and a very vast... 

Transition rules are applied in parallel to all cells, so the state to be adopted by each cell in the next cycle is determined before any of them changes its current state. Because of this feature, CA models are appropriate for the description of processes in which events occur simultaneously and largely independently in different parts of the system and, like several other AI algorithms, are well-suited to implementation on parallel processors. Although the states of the cells change as the model runs, the rules themselves remain unchanged throughout a simulation. 

Steps 2 and 3, in which the environment plays no direct part, can be run independently for each prototype rule in the CS thus the system is well suited to implementation on parallel processor machines. In the next section, we consider the components that form the system in more detail. 

Hartmann et al. (2006) reported very detailed simulation results (see also Hartmann, 2005) (Fig. 9). Their LB simulation was restricted to a lab-scale vessel 10 L in size only for which 2403 lattice cells a bit smaller than 1 mm2 were used the temporal resolution was 25 ps only. A set of 7 million mono-disperse spherical particles 0.3 mm in size was released in the upper 10% part of the vessel. At the moment of release, the local volume fraction amounted to 10%. The particle properties were those of calcium chloride. The simulation was carried out on 30 parallel processors of an SGI Altrix 3700 system and required for 6 weeks for 100 impeller revolutions. 

In working through process control examples, we found that many calculations, data checks, rate checks and other computationally intensive tasks are done at the first level of inference. Considerations of computational efficiency led to a design utilizing two parallel processors with a shared memory (Figure 1). One of the processors is a 68010 programmed in C code. This processor performs computationally intensive, low level tasks which are directed by the expert system in the LISP processor. 

Figure 1. Design for the LMI system for process control using two parallel processors with a shared memory.

For a polyatomic reactant with many degrees of freedom the numerical calculations required to execute the program outlined above can easily achieve a scale that is impossible to handle even with a vectorized parallel processor supercomputer. The simplest approximation that reduces the scale of the numerical calculations is the neglect of some subset of the internal molecular motions, but this approximation usually leads to considerable error. A more sophisticated and intuitively reasonable approximation [72, 73] is to reduce the system dimensionality by placing constraints on the values of the internal molecular coordinates (instead of omitting them from the analysis). 

The system was equilibrated for more than 100 ps the equilibration was checked by monitoring energy trajectories and root mean square (rms) deviations from the X-ray structure. Using 8 parallel processors, it took 1 hour to calculate a 2 ps simulation. The restraints on the system during the molecular dynamics calculation were released in a stepwise mode, first on side chains and then on other atoms. The rms deviation of all Ca atoms between the cytochrome b subimit in the X-ray structure and the unrestrained model after 120 ps was 1.38 A. 

Much of the pioneering work in applying parallel processing to large problems in computational chemistry is due to Enrico Clementi and coworkers, and the development of the LCAP parallel processing systems (see Ref. 90 and the numerous references therein). The LCAP (loosely coupled array of processors) project commenced in 1983, with the stated aim of coupling readily available commercial processors to form a system that is not massively parallel, but rather is modular and can be expanded to match the degree of parallelism that a set of applications can support. 

The development and efficient implementation of a parallel direct SCF Hartree-Fock algorithm, with gradients and random phase approximation solutions, are described by Feyereisen and Kendall, who discussed details of the structure of the parallel version of DISCO. Preliminary results for calculations using the Intel-Delta parallel computer system were reported. The data showed that the algorithms were efficiently parallelized and that throughput of a one-processor Cray X-MP was reached with about 16 nodes on the Intel-Delta. The data also indicated that sequential code, which was not a bottleneck on traditional supercomputers, became time-critical on parallel computers. 

The KSR/Series consists of the KSRl and more recent KSR2. These MIMD parallel MPP systems have globally addressable virtual shared memory. The hierarchical memory has strong coherency and is patented under the name ALLCACHE. A cell is made from Kendall Square s custom and proprietary RISC processor and 32 Mbyte of ALLCACHE memory. The 64-bit KSR2... 

Products/technologies The company sells the CombiTec parallel synthesis system, an organic chemical synthesizer that includes a robotic sample processor, and reaction blocks of 8-56 chambers. Other products include the TRAC system for high-throughput screening with more than 100 microplates the GENESIS Series Robotic Sample Processor (RSP) fully automated microplate-based systems and the Cavro RSP 9000 Robotic Sample Processor (XYZ module). 

Microelectronics Electrochemical phenomena are essential in the manufacture of electronic and photonic systems as well as responsible for the quality and reliability of such systems. Applications and research are outlined in areas that include manufacture of microcircuits, interconnecting networks, lightwave communication devices, parallel processors, content-addressable memories, and nerve-electronic interfaces. 

Because multilayer interconnecting networks are an important element of advanced chips and parallel processors, it is essential that an understanding of the corrosion processes that affect their reliability be developed. Needed are methods to quantify metal corrosion and ion transport in polymers and means to identify electrochemically reliable metal-polymer systems. 

Chemical Network Problems Solved on NASA/Goddard s Massively Parallel Processor (MPP) Computer Symbolic Computation and Chemical Engineering Model Development An Overall Strategy for the Systematic Synthesis and Optimization of Large-Scale Chemical Processing Systems Synthesis of Solids Processing (Heuristics Selection)... 

Given these characteristics, it is evident that large-scale semiempirical SCF-MO calculations are ideally suited for vectorization and shared-memory parallelization the dominant matrix multiplications can be performed very efficiently by BLAS library routines, and the remaining minor tasks of integral evaluation and Fock matrix construction can also be handled well on parallel vector processors with shared memory (see Ref. [43] for further details). The situation is less advantageous for massively parallel (MP) systems with distributed memory. In recent years, several groups have reported on the hne-grained parallelization of their semiempirical SCF-MO codes on MP hardware [76-79], but satisfactory overall speedups are normally obtained only for relatively small numbers of nodes (see Ref. [43] for further details). 

Meinhart, C.D., Prasad, A.K., and Adrian, R.J., A parallel digital processor system for particle image velocimetry, Meas. Sd. Technol., 4, 619-626 (1993). 

It is readily seen that the systematic grid search rapidly leads to a combinatorial block. For example, =4 and Acf 10° resrrlts in the generation of N=36 1.6 x 10 conformations. If this brate-force approach may still be feasible under the rigid body rotation (non-adiabatic) approximation, it rapidly becomes beyond the capacity of present-day computers when full geometry relaxation (adiabatic) is employed for each conformation generated. In this partictdar application comfort may come from coarse-grained parallelized corrrputer systems where many hundreds or even thousands of processors may share the computing burden. 

In the area of computer hardware and operating systems the chemical/pharmaceutical industry increased its use of Digital Equipment Corporation (DEC) VAX computers with the VMS operating system, increased its use of IBM PC computers, and began to talk about more exotic workstations and parallel processors. 

Back-end Processors. An increasing number of hardware vendors supply back-end and parallel processors. These processors usually connect to mini/mainframe computers and increase the throughput of the system for critical operations. Examples are back-end processors that serve as database machines and parallel processors for use with certain numeric-computation algorithms. Many of these processors use the UNIX operating system, so programs common to the chemical/pharmaceutical industry must be converted for use on them. 

The current system mode of a healthy system can be identified by evaluating all ARRs for all feasible switch state combinations. This may require considerable computational effort. The computational time, however, can be reduced by distributing the task on multiple parallel processors. Moreover, this task is only necessary if an initial system mode is not known or if the last known system mode is no longer valid because rapid system modes have taken place while system mode identification is still in progress. Once the current system mode is known, the all-mode FSM can be consulted to identify a subset of ARRs that is to be evaluated to identify the new current system. This has been illustrated by application to a simple circuit with three semiconductor switches. 

A future technique may be an optical parallel processor such as a TSE computer this idea is illustrated in Fig. 24. Some functional devices such as optical AND and OR elements based on semiconductor materials must be developed to obtain this sophisticated system. The two-dimensional configuration made of planar microlenses and a surface-emitting laser array will be very helpful. 

Parallel database systems consist of multiple processors and multiple disks which are part of a fast interconnection network. 

In practical micro-reactors designed for fuel processor systems the channel dimensions are in the range 250-1000 gm in height and from 250 gm to several millimeters in width. Practical micro-reactors are comprised of a multitude, normally thousands, of channels of identical dimensions which are operated in parallel. Owing to the high number and relatively large dimensions of the chaimels, the pressure drop is in the order of a few millibars to a few tens of millibars. 

---