MPI Communication Library Research Articles

Parallel Environment with High Accuracy for Resolution of Numerical Problems

In this paper we describe a high performance environment, like cluster computers, with high accuracy obtained by use of C-XSC library. The C-XSC library is a (free) C++ class library for scientific computing for the development of numerical algorithms delivering highly accurate and automatically verified results by use of the interval arithmetic. These calculus in high accuracy must be available for some basic arithmetic operations, mainly the operations that accomplish the summation and dot product. Because of these aspects, we wish to use the high performance through a cluster environment where we have several nodes executing tasks or calculus. The communication will be done by message passing using the MPI communication library. To obtain the high accuracy in this environment extensions or changes in the parallel programs had done to guarantee that the quality of final result done on cluster, where several nodes collaborate for the final result of the calculus, maintain the same result quality obtained in one sequential high accuracy environment. To validate the environment developed in this work we done basic tests about the dot product, the matrix multiplications, the implementation of interval solvers for banded and dense matrices and the implementation of some numeric methods to solve linear systems with the high accuracy characteristic (some of the methods implemented are used in real life applications like hydrodynamic, agriculture and power electric systems). With these tests we done analysis and comparisons about the performance and accuracy obtained with and without the use of C-XSC library in sequential and parallel programs. With the implementation of these routines and methods will be open a large research field about the study of real life applications that need during their resolution (or in part of their resolution) to calculate arithmetic operations with more accuracy than the accuracy obtained by the traditional computational tools. Our software run on labtec (UFRGS) and Colorado (UPF) clusters. Nowadays we are working in the implementation of parallel versions of programs to solve linear systems (without and with high accuracy) and the optimization of C-XSC library on cluster computers.

Atmospheric dispersion and chemical pollutant transformation simulated with parallel calculations using two PC clusters

The numerical simulation of atmospheric pollution needs long calculation times, mainly due to the chemistry involved, for example in photochemical episodes for which a lot of species and reactions must be taken into account. Parallelisation can reduce these calculation times. In this work, we have made parallel computation with a Eulerian transport/chemical model with the MPI communication library. We used two PC clusters, each PC having similar CPU and memory capacity but being connected in a network with different management systems. Three cases have been simulated, an inert one, one with a simple chemical scheme (nine species and nine reactions) and the last one with a detailed chemistry mechanism (121 reactions and 63 species). With no chemistry, the efficiency falls rapidly due to communication between neighbouring blocks. With chemistry, however, the efficiency is better, and we note that the greater the number of species taken into account the higher the efficiency. This is easily explained by the absence of messages passing in chemistry parts of the program. The present simulations are representative of atmospheric pollution episodes. The efficiency for calculations with photochemistry and using eight runs is about 50%, which corresponds to a four-fold increase in speed, which is interesting only for long-term applications.

Solving large FPT problems on coarse-grained parallel machines

Fixed-parameter tractability (FPT) techniques have recently been successful in solving NP-complete problem instances of practical importance which were too large to be solved with previous methods. In this paper, we show how to enhance this approach through the addition of parallelism, thereby allowing even larger problem instances to be solved in practice. More precisely, we demonstrate the potential of parallelism when applied to the bounded-tree search phase of FPT algorithms. We apply our methodology to the k-V ertex C over problem which has important applications in, for example, the analysis of multiple sequence alignments for computational biochemistry. We have implemented our parallel FPT method for the k-V ertex C over problem using C and the MPI communication library, and tested it on a 32-node Beowulf cluster. This is the first experimental examination of parallel FPT techniques. As part of our experiments, we solved larger instances of k-V ertex C over than in any previously reported implementations. For example, our code can solve problem instances with k⩾400 in less than 1.5 h .

SKaMPI: A Comprehensive Benchmark for Public Benchmarking of MPI

The main objective of the MPI communication library is to enable portable parallel programming with high performance within the message‐passing paradigm. Since the MPI standard has no associated performance model, and makes no performance guarantees, comprehensive, detailed and accurate performance figures for different hardware platforms and MPI implementations are important for the application programmer, both for understanding and possibly improving the behavior of a given program on a given platform, as well as for assuring a degree of predictable behavior when switching to another hardware platform and/or MPI implementation. We term this latter goal performance portability, and address the problem of attaining performance portability by benchmarking. We describe the SKaMPI benchmark which covers a large fraction of MPI, and incorporates well‐accepted mechanisms for ensuring accuracy and reliability. SKaMPI is distinguished among other MPI benchmarks by an effort to maintain a public performance database with performance data from different hardware platforms and MPI implementations.

Hardware-Assisted Characterization of NAS Benchmarks

The UAH Logging, Trace Recording, and Analysis instrumentation (ULTRA) provides highly repeatable (0.0002% variation) application instruction counts for parallel programs which are invariant to the communication network used, the number of processors used, and the MPI communication library used. ULTRA, implemented as an MPI profiling wrapper, avoids the data collection system artifacts of time-based measurements by using instruction counts as the basic measure of work performed and records the operation performed and the amount of data sent for each network operation. These measurements can be scaled appropriately for various target architectures. ULTRA's instrumentation overhead is minimized by using the Pentium II processors's performance monitoring hardware, allowing large, production-run applications to be quickly characterized. Traces of the NAS benchmarks representing 6.67×1012 application instructions were generated by ULTRA. The application instructions executed per byte injected into the network and the instructions executed per message sent were computed from the traces. These values can be scaled by the expected processor performance to estimate the minimum network performance required to support the programs. It is impossible to use time-based measurements for this purpose due to measurement artifacts caused by the background processes and the communication network of the data collection system.

Cell-vertex Based Parallel and Adaptive Explicit 3D Flow Solution on Unstructured Grids

A parallel adaptive Euler flow solution algorithm is developed for 3D applications on distributed memory computers. Significant contribution of this research is the development and implementation of a parallel grid adaptation scheme together with an explicit cell vertex-based finite volume 3D flow solver on unstructured tetrahedral grids. Parallel adaptation of grids is based on grid-regeneration philosophy by using an existing serial grid generation program. Then, a general partitioner repartitions the grid. An adaptive sensor value, which is a measure to refine or coarsen grids, is calculated considering the pressure gradients in all partitioned blocks of grids. The parallel performance of the present study was tested. Parallel computations were performed on Unix workstations and a Linux cluster using MPI communication library. The present results show that overall adaptation scheme developed in this study is applicable to any pair of a flow solver and grid generator with affordable cost. It is also proved that parallel adaptation is necessary for accurate and efficient flow solutions.

A new ignition model for spark-ignited engine simulations

The amount of spark energy deposited into the combustion chamber is key to an optimum ignition as one can end up with misfires when this energy is low or with other undesired effects on engine performance and byproducts when it is high. Experimentally, up to now, no one has been able to correlate the combustion outcome accurately to the spark parameters in a controllable way. Theoretical investigation and computer modeling is leading to a better understanding of how spark flames propagate. A new computational approach to ignition dynamics is presented here for spark-ignited (SI) engine combustion simulations. Our computational model, using the MPI communication library, attempts to solve temporal and spatial equations of the electromagnetic (EM) equations in conjunction with the well-known Navier-Stokes equations of the standard KIVA-3 engine code. The interaction between the gas and the flame (plasma) kernel in the spark region is computed through the momentum and energy exchange between these two fields. Preliminary results show a distinct spatial distribution of physical quantities at the flame front and within the inflammation zone. A slight change in the spark discharge current has significant impact on the combustion and emissions. Enhanced accuracy of spark ignition modeling might help us better compute the early flame propagation and its influence on the cyclic variability of engines, potentially leading to design of new spark plugs.

PARALLEL INCOMPLETE LU FACTORIZATION AS A PRECONDITIONER FOR KRYLOV SUBSPACE METHODS

In this paper we describe a new method for the ILU(0) factorization of sparse systems in distributed memory multiprocessor architectures. This method uses a symbolic reordering technique, so the final system can be grouped in blocks where the rows are independent and the factorization of these entries can be carried out in parallel. The parallel ILU(0) factorization has been tested on the Cray T3E multicomputer using the MPI communication library. The performance was analysed using matrices from the Harwell–Boeing collection.

Modelling coastal sediment transport on a parallel computer

The movement of suspended sediments in estuaries and coastal environments cause many problems, e.g. siltation of navigation waterways, harbours and coastal environments. The transport of sediment in a tidal environment is rather complex. This is due to the complexity of the coastal hydrodynamics and the interaction of the sediments with the flow. Numerical simulations of the tidal flow and sediment transport are useful tools for planning, design and management. Computations become expensive when the flow model and the sediment transport model are coupled, especially when the computations are done with small grid sizes. In this paper we describe a coupled model for the Belgian coast area, where a grid size of 75 m ∗ 75 m is needed to resolve the complicated bathymetry. In order to improve the performance of the coupled model, the codes have been modified to run on a parallel computer (IBM/SP2). The modification includes a transformation from FORTRAN 77 code into FORTRAN 90. The MPI communication library is used for exchanging information between the processors. The model results and the performance on the IBM/SP2 machine are presented and discussed.

Optimizations for Efficient Array Redistribution on Distributed Memory Multicomputers

Appropriate data distribution has been found to be critical for obtaining good performance on distributed memory multicomputers such as the Thinking Machines CM-5, Intel Paragon, and IBM SP-1/SP-2. It has also been found that some programs need to change their distributions during execution for better performance (redistribution). This work focuses on automatically generating efficient routines for redistribution. We present a new mathematical representation for regular distributions called FALLS and then discuss algorithms for redistribution based on this representation. One of the significant contributions of this work is being able to handle arbitrary source and target processor sets while performing redistribution. Another important contribution is the ability to handle an arbitrary number of dimensions for the array involved in the redistribution in a scalable manner. Our implementation of these techniques is based on the MPI communication library. The results presented show the efficiency and scalability of our redistribution algorithm.