A. J. Field

Parallel Programming Using Skeleton Functions

Programming parallel machines is notoriously difficult. Factors contributing to this difficulty include the complexity of concurrency, the effect of resource allocation on performance and the current diversity of parallel machine models. The net result is that effective portability, which depends crucially on the predictability of performance, has been lost. Functional programming languages have been put forward as solutions to these problems, because of the availability of implicit parallelism. However, performance will be generally poor unless the issue of resource allocation is addressed explicitly, diminishing the advantage of using a functional language in the first place.

Mighty small: Observing and modeling individual microbes becomes big science

Progress in microbiology has always been driven by technological advances, ever since Antonie van Leeuwenhoek discovered bacteria by making an improved compound microscope. However, until very recently we have not been able to identify microbes and record their mostly invisible activities, such as nutrient consumption or toxin production on the level of the single cell, not even in the laboratory. This is now changing with the rapid rise of exciting new technologies for single-cell microbiology (1, 2), which enable microbiologists to do what plant and animal ecologists have been doing for a long time: observe who does what, when, where, and next to whom. Single cells taken from the environment can be identified and even their genomes sequenced. Ex situ, their size, elemental, and biochemical composition, as well as other characteristics can be measured with high-throughput and cells sorted accordingly. Even better, individual microbes can be observed in situ with a range of novel microscopic and spectroscopic methods, enabling localization, identification, or functional characterization of cells in a natural sample, combined with detecting uptake of labeled compounds. Alternatively, they can be placed into fabricated microfluidic environments, where they can be positioned, exposed to stimuli, monitored, and their interactions controlled “in microfluido.” By introducing genetically engineered reporter cells into a fabricated landscape or a microcosm taken from nature, their reproductive success or activity can be followed, or their sensing of their local environment recorded.

A simple recursive tessellator for adaptive surface triangulation

Abstract Sometimes there is a need to create a triangular mesh approximation of a parametric surface. If the parameterization is nonuniform (compressed in some areas, stretchy in others), a uniform grid in parameter space becomes distorted and provides a bad approximation. We describe how to create an adaptive triangulation of such a surface, provided the user of the algorithm provides a routine split_edge() which indicates whether a particular edge is close enough to the surface or requires splitting, and optionally a routine flat_enough() which t ells whether a triangle whose edges appear adequate is indeed flat enough, or requires further subdivision. Our contribution is a simple algorithm for guaranteeing that the topology of the resulting mesh is well formed in the sense that there ar e no cracks between triangles (i.e., T-junctions), and for ensuring that subdivision halts at a given point. There is also rudimentary support from trimmed surfaces. Source code for the algorithm is available online.

Changing the marks-based culture of learning through peer-assisted tutorials

We describe and evaluate an approach to student learning that aims to instil a culture of formative assessment based on peer-assisted learning. The idea is for suitably qualified undergraduates to assist in the running of weekly first-year tutorials. They mark submitted work, provide written and verbal feedback and lead problem-solving discussions during tutorials. However, contrary to normal practice, the marks they award do not contribute to the students’ end-of-year total; all tutorial work becomes essentially voluntary. We report results from a pilot implementation of the scheme over a 12 month period in an engineering department at a leading academic institution. The scheme was such that a comparative and triangulated assessment was possible among the students and tutor team. Results show no discernible degradation in student attendance, submission rates and performance in either the weekly exercises or end-of-year examinations. Important benefits to the peer tutors are also found.

M-Tree: A Parallel Abstract Data Type for Block-Irregular Adaptive Applictions

This paper describes an abstract data type called M-Tree — a generalization of a quadtree which captures both the data structure and computational structure common to many adaptive problems in science and engineering. It is equipped with a rich set of access functions including higher-order operators describing commonly used computational patterns in parallel adaptive computations. This provides a uniform high level abstraction of a wide range of applications including adaptive mesh refinement and adaptive particle simulation and thus enables such applications to be constructed systematically and efficiently. We present examples in which an M-tree is used to solve both an adaptive heat-flow problem and N-body particle simulation. The structured abstraction of commonly-occurring computation patterns in the application provides us with the opportunity to investigate various approaches to load balancing and communication minimization using caching and other techniques. These optimizations are applicable to other problems with a similar structure.

Overcoming barriers to restructuring in a modular visualisation environment

This paper explores the potential for automatic cross-component optimisation in the Python / VTK-based MayaVi modular visualisation environment. The idea is to delay execution of the VTK components called from the MayaVi tool, which requires no significant structural change to the MayaVi code base, but which opens up the possibility for dynamic performance optimisations such as tiling, fusion, memoisation and shared-memory parallelisation. The paper concludes with experimental results on an unstructured mesh hierarchy model from an adaptive three-dimensional gravity current simulation.

Response Times in Client-Server Systems

Response time is the key performance measure in on-line transaction processing systems and other client-server architectures. Not only is it important to achieve a low average value and correspondingly high throughput, but response time should also be fairly consistent in order to provide a good quality of service. We develop a new algorithm for computing the probability density function of response times in Markovian models of client-server systems. We model the clients and servers by central server queueing networks and obtain response time densities as simple functions of time under independence assumptions that are shown to hold asymptotically as network size increases. The communication network is modelled as a single server queue with mean service time determined by its operational characteristics. We consider an Ethernet and construct a new model, of interest in its own right, that captures details not modelled hitherto. This model is validated against simulation and shows good agreement up to moderate utilisations, the normal operating environment for Ethernets. The whole client-server model is implemented in the Metron Athene Client-Server capacity planning tool and sample runs are examined.

Performance modelling of parallel computer architectures

In this paper we describe two types of complex server aggregations which can be used to model collections of components in certain types of parallel computer systems and give a case study showing how the aggregations may be applied in practice. Analytical models of such systems are becoming increasingly important as a means of guiding the often complex design processes, particularly since recent developments in VLSI technology now make it possible to fabricate many paper-designs hitherto impractical for reasons of cost. We argue that aggregations of the type described are essential in the modelling of parallel systems; using the proposed techniques, large numbers of components can be modelled as queue-length-dependent servers within a queueing network in which the number of servers is the same as the number of distinct types of processing element in the system being modelled. Because the number of severs in the model is fixed i.e. is independent of the number of processors, very large multiprocessor systems can be modelled efficiently with no explosion in the size of the state space.

Sojourn times in a random queue with and without preemption

Abstract We consider an equilibrium M/M/1 queue in which customers arrive at a random position, both for the preemptive and non-preemptive cases. The Laplace–Stieltjes transforms of their sojourn time distributions are derived and in the latter case it is verified that the result is that of an M/M/1 queue where the server chooses the next task to serve at random.

Tackling Complexity in High Performance Computing Applications

We present a software framework that supports the specification of user-definable configuration options in HPC applications independently of the application code itself. Such options include model parameter values, the selection of numerical algorithm, target platform etc. and additional constraints that prevent invalid combinations of options from being made. Such constraints, which are capable of describing complex cross-domain dependencies, are often crucial to the correct functioning of the application and are typically either completely absent from the code or a hard to recover from it. The framework uses a combination of functional workflows and constraint solvers. Application workflows are built from a combination of functional components: higher-order co-ordination forms and first-order data processing components which can be either concrete or abstract, i.e. without a specified implementation at the outset. A repository provides alternative implementations for these abstract components. A constraint solver, written in Prolog, guides a user in making valid choices of parameters, implementations, machines etc. for any given context. Partial designs can be stored and shared providing a systematic means of handling application use and maintenance. We describe our methodology and illustrate its application in two classes of application: a data intensive commercial video transcoding example and a numerically intensive incompressible Navier–Stokes solver.