We address the problem of data parallel processing for computational quantum chemistry (CQC). CQC is a computationally demanding tool to study the electronic structure of molecules. An important algorithmic component of these computations is the evaluation of Electron Repulsion Integrals (ERIs). A key problem with ERI evaluation is controlflow variation between different ERI evaluations, which can only be resolved at runtime. This causes the computation to be unsuitable for data parallel execution. However, it is observed that although there is variation between ERI evaluations, the variation is limited; in fact there are a limited number of ERI classes present within any given workload. Conceptually, it is possible to classify the ERIs into sizable sets, and execute these sets in a data parallel fashion. Practically, creating these sets is computationally expensive. We describe an architecture to perform this thread sorting, where high throughput is achieved with small associative and multiport memories. The performance of the prototype is evaluated with FPGA synthesis. We go on to envision other uses for thread sorting, in general-purpose manycore architectures.

Most UAVs (unmanned aerial vehicles) today use the GPS (Global Positioning System) for navigation. While GPS is an accurate and usually reliable aid to navigation, it can be unavailable for various reasons. The aim of this project is to develop an alternative navigation method, which can be used when GPS transmissions are not available. The method suggested here is called feature based navigation. It works by comparing the current image taken by the on-board camera with previous reference images, usually the most recent, obtained when GPS was available. The system then compares locations of similar features extracted from the images using these to determine the location of UAV. We outline the results of the method using real data taken from the data log of a UAV in flight

The unconventional air frames used by many small fixed-wing unmanned aerial vehicles (UAVs) pose difficulties in determining their dynamic models accurately. This is further complicated by the low Reynolds numbers flight regimes encountered by these aircraft. This paper studies the attitude flight dynamics identification of an unconventional model-scale UAV which has only two independently actuated elevon surfaces (no rudder or elevators). The study presents the first steps towards the development of autopilots for this kind of UAV. Utilizing the real-time flight data collected from a human-controlled test flight, the standard identification approaches were applied to obtain a MIMO linear model with given configuration deduced from a theoretical study. The simulation-based validation of a simplified model was undertaken which shows acceptable modelling accuracy. Further flight tests controlled by an autopilot tuned according to the model are currently being undertaken

There are a number of approaches to the application of subspace techniques for solving spectral estimation problems. These approaches are derived from the covariance matrix which is constructed from incoming data. The covariance matrix can be broken down through the use of appropriate matrix properties and eigen-decomposition techniques into two subspaces. The performance of three traditional algorithms which incorporate subspace techniques in direction of arrival are evaluated under both white and 1/f noise conditions. 1/f noise is chosen because it is typical of the EEG signals. Simulation results suggest that the Johnson and DeGraaf (1982) direction finding algorithm performs best under both noise environments. A typical sample of EEG data was used to evaluate the performance of the three algorithms. The Johnson and DeGraaf algorithm gives estimates for the direction of the signal which approximately agree with the anatomical locations of possible electrocortical generators.>

The SDEM code models systems of interacting blocks of rock using the distinct element (DE) method, which represents these systems as discontinuums with each block acting under Newton's laws of motion. The data structures associated with the DE method make the task of obtaining performance gains through vectorisation difficult. Typical systems, however, contain thousands of blocks and there is the potential to perform calculations associated with groups of blocks in parallel.

Numerical Weather Prediction (NWP) is acknowledged as being of vital importance to economy. The demand that NWP places on computing system performance has increased dramatically since the introduction of computer systems and is still growing. This paper describes the implementation of a one-level barotropic spectral NWP model using the high level functional language SISAL. Our initial approach was a direct transliteration from FORTRAN to SISAL, the result of which was a large amount of array copying and loss of parallelism. The algorithm of the model was then rearranged to enable effective exploitation of parallelism by an optimizing SISAL compiler. The parallel implementation technique and its potential for modeling with larger model sizes are discussed, and the improved results for an Encore Multimax multiprocessor are presented.

This paper examines the two convention models of data-flow machines, static and dynamic. The advantages and disadvantages of each scheme are described. A hybrid architecture is presented which gains the advantages of both. This architecture has been incorporated into a new data-flow machine built at the Royal Melbourne Institute of Technology. Some simulations are presented which illustrate the advantages of the hybrid approach.