High Performance Computing (HPC) is the key to solving many scientific, financial, and engineering problems. Computer clusters are now the dominant architecture for HPC. The scale of clusters, both in terms of processor per node and the number of nodes, is increasing rapidly, reaching petascales these days and soon to exascales. Inter-process communication plays a significant role in the overall performance of HPC applications. With the continuous enhancements in interconnection technologies and node architectures, the Message Passing Interface (MPI) needs to be improved to effectively utilize the modern technologies for higher performance. After providing a background, I present a deep analysis of the user level and MPI libraries over modern cluster interconnects: InfiniBand, iWARP Ethernet, and Myrinet. Using novel techniques, I assess characteristics such as overlap and communication progress ability, buffer reuse effect on latency, and multiple-connection scalability. The outcome highlights some of the inefficiencies that exist in the communication libraries. To improve communication progress and overlap in large message transfers, a method is proposed which uses speculative communication to overlap communication with computation in the MPI Rendezvous protocol. The results show up to 100% communication progress and more than 80% overlap ability over iWARP Ethernet. An adaptation mechanism is employed to avoid overhead on applications that do not benefit from the method due to their timing specifications. To reduce MPI communication latency, I have proposed a technique that exploits the application buffer reuse characteristics for small messages and eliminates the sender-side copy in both two-sided and one-sided MPI small message transfer protocols. The implementation over InfiniBand improves small message latency up to 20%. The implementation adaptively falls back to the current method if the application does not benefit from the proposed technique. Finally, to improve scalability of MPI applications on ultra-scale clusters, I have proposed an extension to the current iWARP standard. The extension improves performance and memory usage for large-scale clusters. The extension equips Ethernet with an efficient zero-copy, connection-less datagram transport. The software-level evaluation shows more than 40% performance benefits and 30% memory usage reduction for MPI applications on a 64-core cluster. / Thesis (Ph.D, Electrical & Computer Engineering) -- Queen's University, 2010-10-16 12:25:18.388

No description available.

Selenium has been shown to have multiple biochemical effects ranging from nutrient deficiency at low levels to toxicity at high levels. This duality of concern has led to a demand for increased numbers of highly accurate and precise determinations of selenium in biological materials. A convenient procedure was developed for determining selenoamino acids by HPLC-THG-AAS, based on the derivatization of these analytes with Sanger's reagent. Selenomethionine, selenocystine and selenocysteine (after blocking the free selenol group with phenylmercurio cation) were converted to their N-2,4-dinitrophenyl derivatives, and separated on a Nucleosil 5-NO$ sb2$ column with methanolic mobile phase containing acetic acid and triethylamine. Furthermore, an improved HPLC-AAS interface design was modified and optimized for the detection of selenium in HPLC column eluate. The new design was (i) compatible with aqueous mobile phases containing volatile buffers and (ii) provided equivalent molar response to analytes containing Se($-$II), Se(+IV) and Se(+VI). A method for simultaneously determination of selenate, selenite, selenocystine, selenomethionine and selenoethionine was developed by using the HPLC-AAS system with aqueous acetic acid containing ammonium acetate as eluate solution on the cyanopropyl column. The equivalent low ng limits of detection (1-2 ng as Se) for different oxidation states of selenium analytes were obtained using several different mobile phases and/or columns. A phenol extraction procedure for selenate, selenite, selenocystine, selenomethionine and selenoethionine was evaluated for the determination of these selenium analytes in natural waters and wheat samples. The current HPLC-AAS system provides an inexpensive alternative to conventional techniques for the determination of selenium analytes in environmental samples.

Standards of Amadori rearrangement products (ARP) were synthesized for the purpose of developing analytical techniques and performing mechanistic studies related to their thermal decomposition. Several synthetic strategies were explored. An HPLC analytical system with a diode array detector was coupled to a fluorometer and an electrochemical detector, in order to detect simultaneously and on-line, a wide variety of degradation products of ARPs and to follow their kinetics. The potential of such a system to analyze complex Maillard mixtures was demonstrated. The kinetics of the reaction of glucose with morpholine (a Strecker inactive analogue of proline was used in order to simplify the kinetics) to produce Amadori morpholine was studied under experimental conditions that minimize side reactions and maximize Amadori product formation. At specific time intervals, the samples were analyzed for the presence of reactants and Amadori product by the multidetector HPLC system. Color and fluorescence were also measured. The data obtained were used to calculate the rate constants for the formation and degradation of Amadori product. A mechanistic model that statistically fitted the kinetic data was proposed. To further understand the details of the decomposition mechanism of Amadori proline, different mass spectrometric experiment were performed. High resolution, linked-field scan and neutral loss experiments have indicated that 1-((2$ sp prime$-carboxyl)pyrrolidinyl)-1-deoxy- scD-fructose (Proline Amadori product) followed two main pathways of fragmentation under electron impact conditions; one initiated by the ring oxygen and the other by the amino acid nitrogen, producing two well stabilized fragment ions; oxonium and imminium ions. In addition, ortho-elimination reactions initiated by O or N-centered radical sites were shown to produce the most intense peaks and diagnostically important ions for the identification of Amadori products. However this approach can only pro

The need for selective and inexpensive detectors in liquid chromatography is of considerable interest in the determination of sulfur compounds. Of the available-selective sulfur methodologies, flame photometric detector coupled to gas chromatography is the most widely used. It has proven to be a sensitive and selective method for detection of heat stable and volatile sulfur compounds. Fundamentally, this technique is not applicable to high boiling and/or thermolabile sulfur compounds. More recently, hyphenated flame photometric detector has been utilized, with limited success, to monitor sulfur species in liquid chromatography. However, existing HPLC-FPD methodologies have never been applied to real samples, due to the low population of S 2, the emitting species, and the quenching effects of the other species present in the flame. / In this work, two total consumption high-performance liquid chromatography flame photometric (HPLC-FPD) interfaces compatible with either methanolic or aqueous mobile phases are described and optimized for monitoring low volatile and thermally fragile sulfur compounds in biological samples. Each interface was fuelled either by methanol or by hydrogen. The all quartz interfaces enclosed four consecutive thermal processes: (a) thermovaporization of the HPLC effluent; (b) pyrolysis of the organic matrix (including sulfur species) in a kinetic H2/O2 flame; (c) conversion of the oxidized sulfur compounds to H2S in a reducing post-combustion stage fuelled by hydrogen; and (d) transport of the generated hydrides towards a hydrogen radical rich surrounding of an inverted hydrogen-oxygen diffusion flame. Chemiluminescence induced in the last step was integrated as a narrow beam in a light-guide positioned remotely from the analytical cool flame and oriented towards a photomultiplier unit. Radioisotopic assays demonstrated that sulfur (as H235SO4) was transferred quantitatively to the analytical flame. Indirect evidence suggested that sulfur was hydrogenated in the post-combustion step via a thermochemical hydride generation process to mediate the formation of S2. The linearity of calibration graphs (0.9950 < r < 0.9986), where r is the correlation coefficient) and unprecedented HPLC-FPD limits of detection for sulfur compounds (1.5 etag/s for 2-methylthiophene, 2.25 etag/s for carbon disulfide, and 4.5 etag/s for ethanesulfonic acid) allowed for the speciation of sulfur species in garlic extracts. Alternatively, modification of the methanol fuelled interface to a hydrogen fuelled reactor allowed detection of thiosulfinates and high molecular weight sulfur compounds in horse kidney and garlic extracts, respectively.

The design and construction of an electrochemical detector with two working electrodes located on the opposite walls of a thin-layer cell and its use as a detector for High Pressure Liquid Chromatography (HPLC) in the analysis of catecholamines in human cerebrospinal fluid are described. The location of the electrodes in this manner permits an electrochemically reversible or quasireversible couple to be electrolized more than once as it passes through the detector. If one electrode is held at a potential where oxidation takes place and the second electrode at a potential where reduction of this oxidized form back to the starting material occurs, then the current produced increases proportionately to the number of conversions that take place. A comparison of this cell in the dual-working-electrode and single-working-electrode mode shows an improvement in the signal-to-noise ratio by a factor of six. This HPLC system with electrochemical detection has been used for the first time to detect norephinephrine (141 pg/mL) and dopamine (262 pg/mL) in human cerebrospinal fluid.

This dissertation provides a comprehensive overview of the theory and practice of Dynamic Voltage/Frequency Scaling (DVFS) in the High Performance Computing (HPC) environment. We summarize the overall problem as follows: how can the same level of computational performance be achieved using less electrical power? Equivalently, how can computational performance be increased using the same amount of electrical power? In this dissertation we present performance and architecture models of DVFS as well as the Adagio runtime system. The performance model recasts the question as an optimization problem that we solve using linear programming, thus establishing a bound on potential energy savings. The architectural model provides a low-level explanation of how memory bus and CPU clock frequencies interact to determine execution time. Using insights provided from these models, we have designed and implemented the Adagio runtime system. This system realizes near-optimal energy savings on real-world scientific applications without the use of training runs or source code modification, and under the constraint that only negligible delay will be tolerated by the user. This work has opened up several new avenues of research, and we conclude by enumerating these.

No description available.

No description available.

As the demands of computational science and engineering simulations increase, the size and capabilities of High Performance Computing (HPC) clusters are also expected to grow. Consequently, the software providing the application programming abstractions for the clusters must adapt to meet these demands. Specifically, the increased cost of interprocessor synchronization and communication in larger systems must be accommodated. Non-blocking operations that allow communication latency to be hidden by overlapping it with computation have been proposed to mitigate this problem. In this work, we investigate offloading a portion of the communication processing to dedicated hardware in order to support communication/computation overlap efficiently. We work with the Message Passing Interface (MPI), the de facto standard for parallel programming in HPC environments. We investigate both point-to-point non-blocking communication and collective operations; our work with collectives focuses on the allgather operation. We develop designs for both flat and hierarchical cluster topologies and examine both eager and rendezvous communication protocols. We also develop a generalized primitive operation with the aim of simplifying further research into non-blocking collectives. We propose a new algorithm for the non-blocking allgather collective and implement it using this primitive. The algorithm has constant resource usage even when executing multiple operations simultaneously. We implemented these designs using CORE-Direct offloading support in Mellanox InfiniBand adapters. We present an evaluation of the designs using microbenchmarks and an application kernel that shows that offloaded non-blocking communication operations can provide latency that is comparable to that of their blocking counterparts while allowing most of the duration of the communication to be overlapped with computation and remaining resilient to process arrival and scheduling variations. / Thesis (Master, Electrical & Computer Engineering) -- Queen's University, 2014-05-29 11:55:53.87