Parallel architectures

Notes and Comments. Further improvements in efficiency were achieved by implementing the method on computers with highly parallel architecture. SISM performs in parallel as LFV which means the speed up Is gained due to longer time stop wliidi cun be u.sed by SISM [20]. 

It can be said that these three main strategies have been applied equally and very often in combination. Basically, the first approach implies the use of a faster computer or a parallel architecture. To some extent it sounds like a brute force approach but the exponential increase of the computer power observed since 1970 has made the hardware solution one of the most popular approaches. The Chemical Abstracts Service (CAS) [10] was among first to use the hardware solution by distributing the CAS database onto several machines. 

Unfortunately, Flynn s classification, although commonly used, is quite restrictive when discussing parallel-architecture computers. There have been several attempts to formulate more detailed classification schemes for the great variety of parallel computers now available. None of these efforts have been entirely successful, and none appear to be in general use. A discussion of representative machines from some of the more common classes follows. 

The observation that certain kinds of parallel-computing architectures best support only certain kinds of problems seems to be general. The further observation that interprocessor communication can be the primary impediment to parallel performance is also general. As of this writing, any hope of a truly general purpose parallel computer seems to be remote. The best hope may He in software efforts that describe problems at higher levels of abstraction, which can then be ported and optimized for different parallel architectures (22). 

This line of research has not lost his momentum. One of the reasons is the eontinuing progress in the computer hardware and software. Methods and algorithms are, and will be, continuously updated to exploit new features made available by eomputer seienee, as for example the parallel architectures, or the neuronal networks, to mention things at present of widespread interest, or even conceptually less significant improvements, as the inerease of fast memory in commereial computers. Computer quantum chemistry is not a mere recipient of progresses in eomputer seienee. Many progresses in the software comes from... 

Carpenter, G. A., and Grossberg, S., A massively parallel architecture for a self-organizing neural pattern recognition machine, Comput. Vis. Graphics Image Process 37,54 (1987b). 

B. Borchers and J. Young, Implementation of a Primal—Dual Method for SDP on a Parallel Architecture, Research Report, 2005. Available at http //infohost.nmt. edu/ borchers/csdp.html. 

The use of computational chemistry to address issues relative to process design was discussed in an article. The need for efficient software for massively parallel architectures was described. Methods to predict the electronic structure of molecules are described for the molecular orbital and density functional theory approaches. Two examples of electronic stracture calculations are given. The first shows that one can now make extremely accurate predictions of the thermochemistry of small molecules if one carefully considers all of the details such as zero-point energies, core-valence corrections, and relativistic corrections. The second example shows how more approximate computational methods, still based on high level electronic structure calculations, can be used to address a complex waste processing problem at a nuclear production facility (Dixon and Feller, 1999). 

We have used genetic programming to evolve a program, which is executed by a processing element of a parallel architecture as shown in Figure 8.7. The processing elements... 

An implementation of the configuration-selecting multireference configuration-interaction method on massively parallel architectures 95... 

Hong, J.W., Studer, V., Hang, G., Anderson, W.F., Quake, S.R., A nanoliter-scale nucleic acid processor with parallel architecture, Nat. Biotechnol. 2004, 22, 435-439. 

Such problem tailoring requires some familiarity with the algorithmic modules. It also demands knowledge of the theoretical and practical strengths and weaknesses of the different minimization methods. With rapidly growing improvements in high-performance super and massively parallel machines,15 16 application-tailored software may be even more important in combination with parallel architectures whose design is motivated by specific applications. 

Hinton, G. E. (1987). Learning translation invariant recognition in a massively parallel network. In PARLE Parallel Architecture and Languages (ed Goos, G. Hartmanis, J.), pp. 1-13. Springer-Verlag, Berlin. 

Each pair of chambers is connected to the protein sample and one of 48 crystallization solution reservoirs. The chip has 480 integrated valves that are actuated through three separately addressable control lines. As a precaution, 48 safety valves are included at the solution inlets to avoid the unwanted loss of protein sample in the unlikely event of an interface valve failure. The remaining two lines simultaneously control all interface and containment valves. By virtue of this parallel architecture and the robustness of the BIM scheme, solutions of varying viscosity, surface tension, pH and ionic strength may be simultaneously metered and mixed at three different mixing ratios using only two hydraulic control lines. 

Fig. 11.5 Parallel architecture of crystallization chip and blow-up of a single unit cell. All unit cells are connected to a central sample port (yellow) and one of 48 unique reagent ports (blue). All interface and containment...

Following a general overview, that summarizes the pioneering role of the LCAP project, o give accounts of work in quantum chemistry, molecular dynamics, and reaction dynamics. Each of these subsections is organized to review the development and current status of codes in the area and to provide an overview of work under way where serial implementations are migrating onto parallel architectures. 

Shared-memory parallel processing was certainly more successful for QC in earlier applications and continues to play a significant role in high performance computational chemistry. A coarse-grained parallel implementation scheme for the direct SCF method by Liithi et al. allowed for a near-asymptotic speed-up involving a very low parallelization overhead without compromising the vector performance of vector-parallel architectures. 

J. R. Savage, Theor. Chim. Acta, 72, 139 (1987). A Parallel Architecture and Programming... 

D. K. Hoffman, O. A. Sharafeddin, D. J. Kouri, M. Carter, N. Nayar, and J. Gustafson, Theor. Chim. Acta, 79, 297 (1991). On the Role of Parallel Architecture Supercomputers in Time-Dependent Approaches to Quantum Scattering. 

---