Due to the massive computational power, Graphics Processing Units (GPUs) have become a popular platform for executing general purpose parallel applications. The majority of on-chip communications in GPU architecture occur between memory controllers and compute cores, thus memory controllers become hot spots and bottle neck when conventional mesh interconnection networks are used. Leveraging this observation, we reduce the network latency and improve throughput by providing a nanophotonic ring network which connects all memory controllers. This new interconnection network employs a new routing algorithm that combines Dimension Ordered Routing (DOR) and nanophotonic ring algorithms. By exploring this new topology, we can achieve to reduce interconnection network latency by 17% on average (up to 32%) and improve IPC by 5% on average (up to 11.5%). We also analyze application characteristics of six CUDA benchmarks on the GPGPU-Sim simulator to obtain better perspective for designing high performance GPU interconnection network.

Evolutionary Structural Optimization (ESO) is a relatively new design tool used to improve and optimise the design of structures. In this method, a few elements of an initial design domain of finite elements are iteratively removed. Such a process is carried out repeatedly until an optimum design is achieved, or until a desired given area or volume is reached. In structural design, there is the demand for the development of design tools and methods that includes optimization. This need is the reason behind the development of methods like Evolutionary Structural Optimization (ESO). It is also this demand that this thesis seeks to satisfy. This thesis develops and examines the program named EVO, with the concept of structural optimization in the ESO process. Taking into account the stiffness and stress constraints, EVO allows a realistic and accurate approach to optimising a model in any given environment. Finally, in verifying the ESO algorithm&rsquo / s and EVO program&rsquo / s usefulness to the practical aspect of design, the work presented herein applies the ESO method to case studies. They concern the optimization of 2-D frames, and the optimization of 3-D spatial frames and beams with the prepared program EVO. Comparisons of these optimised models are then made to those that exist in literature.

Thesis (M. Sc. (Mathematics))--Rhodes University, 2002.

Not available / text

We introduce the naive notion of a relative open book decomposition for contact 3-manifolds with torus boundary. We then use this to construct nice, minimal genus open book decompositions compatible with all of the universally tight contact structures (as well as a few others) on torus-bundles over S¹, following Honda's classification. In an accurate sense, we find Stein fillings of 'half' of the torus bundles. In addition, these give the first examples of open books compatible with the universally tight contact structures on circle bundles over higher genus surfaces, as well, following a pattern introduced by a branched covering of B⁴. Some interesting examples of open books without positive monodromy are emphasized.

- / -

No description available.

- / -

- / -

In his 1947 work "Theory of Braids" Emil Artin asked whether the braid group remained unchanged when one considered classes of braids under linkhomotopy, allowing each strand of a braid to pass through itself but not through other strands. We generalize Artin's question to string links over orientable surface M and show that under link-homotopy surface string links form a group PBn(M), which is isomorphic to a quotient of the surface pure braid group PBn(M). Surface braid groups and their properties are an area of active research by González-Meneses, Paris and Rolfsen, Goçalves and Guaschi, and our work explores the geometric and visual beauty of this subject. We compute a presentation of PBn(M) in terms of the generators and relations and discuss the orderability of the group in the case when the surface in question is a unit disk D.