Portability, predictability and performance for parallel computing: BSP in practice

We report on practical experience using the Oxford BSP Library to parallelize a large electromagnetic code, the British Aerospace finite-difference time-domain code EMMA T:FD3D. The Oxford BS Library is one of the first realizations of the Bulk Synchronous Parallel computational model to be targeted at numerically intensive scientific (typically Fortran) computing. The BAe EMMA code is one of the first large-scale applications to be parallelized using this library, and it is an important demonstration of the cost effectiveness of the BSP approach. We illustrate how BSP cost-modelling techniques can be used to predict and optimize performance for single-source programs across different parallel platforms. We provide predicted and observed performance figures for an industrial-strength, single-source parallel code for a variety of real parallel architectures: shared memory multiprocessors, workstation clusters and massively parallel platforms.

Reviews

[Review of] Lou Charnon-Deutsch, The Spanish Gypsy: The History of a European Obsession, Pennsylvania State University Press: University Park, PA, 2004; 280 pp., 31 illus.; 0271023597, $42.95 (hbk).

Start-up and running forces on bulk solids feeders: experimental findings versus available models

Many different models have been postulated over the years for sizing of feeder drives; these models have different bases, some rationally based and others more rule-of-thumb. Experience of Jenike & Johanson and likewise of The Wolfson Centre in trouble-shooting feeder drives has shown that drive powers are often poorly matched, so there is clearly still some way to go towards establishing a universally-used reliable approach. This paper presents an on-going programme of work designed to measure feeder forces experimentally on a purpose designed testing rig, and to compare these against some of the best known available models, and also against a full size installation. One aspect which is novel is the monitoring of the transition between the “filling stress field” load on the feeder and the “flowing stress field” load.

Comparisons between “sand blast” and “centripetal effect accelerator” type erosion testers

There are two major types of erosion testing devices that are used throughout the world for quantifying particle impact erosion against a solid surface. The first of these uses pressurised air to accelerate abrasive particles through a nozzle so that they impinge upon a target specimen. The second adopts a rotating disc to accelerate abrasive particles using the centripetal effect so that they impinge upon a series of targets arranged around the periphery of the disc. This paper reports the findings of a collaborative project that was designed to compare the performance and results obtained from a rig of each of the two types mentioned above. The sand blast type rig was provided by The Department of Powder Science Technology (POSTEC) at The Telemark Technological Research and Development Centre (TEL-TEK), Porsgrunn, Norway while the centripetal effect accelerator was provided by The Wolfson Centre for Bulk Solids Handling Technology, University of Greenwich, London, UK. The test programme included tests against a wide range of materials that are commonly used in pneumatic handling facilities. (Pneumatic handling is a means of conveying and transporting powders and granular solid materials in bulk in industrial process plant, through pipelines using a gas as the carrier medium.) Olivine sand was used as the abrasive and it was projected against the test specimens at velocities and concentrations commensurate with those seen in pneumatic conveyors. In all instances the materials used in the test programme were taken from the same batch so that scatter of experimental results due to specimen variation was minimised. The paper contains a series of recommendations for erosion testing equipment. A discussion based on the results and their applicability to the prediction of wear in pneumatic conveyors concludes the paper.

The dependence of power consumption of pneumatic conveying systems upon bulk material properties, pipeline bore and routing, and mode of flow

In this paper the dependence of the power consumption of pneumatic conveyors upon conveyed materials, pipeline route and bore, and mode of flow has been examined. The findings are that, with different materials and modes of flow, not only is the amount of power consumed very different but it varies in different ways with pipe bore and routing. Additionally it has been found that, for any given conveying system, the choice of air mover also has a strong influence on the power requirement.