GPRM: a high performance programming framework for manycore processors

Processors with large numbers of cores are becoming commonplace. In order to utilise the available resources in such systems, the programming paradigm has to move towards increased parallelism. However, increased parallelism does not necessarily lead to better performance. Parallel programming models have to provide not only flexible ways of defining parallel tasks, but also efficient methods to manage the created tasks. Moreover, in a general-purpose system, applications residing in the system compete for the shared resources. Thread and task scheduling in such a multiprogrammed multithreaded environment is a significant challenge. In this thesis, we introduce a new task-based parallel reduction model, called the Glasgow Parallel Reduction Machine (GPRM). Our main objective is to provide high performance while maintaining ease of programming. GPRM supports native parallelism; it provides a modular way of expressing parallel tasks and the communication patterns between them. Compiling a GPRM program results in an Intermediate Representation (IR) containing useful information about tasks, their dependencies, as well as the initial mapping information. This compile-time information helps reduce the overhead of runtime task scheduling and is key to high performance. Generally speaking, the granularity and the number of tasks are major factors in achieving high performance. These factors are even more important in the case of GPRM, as it is highly dependent on tasks, rather than threads. We use three basic benchmarks to provide a detailed comparison of GPRM with Intel OpenMP, Cilk Plus, and Threading Building Blocks (TBB) on the Intel Xeon Phi, and with GNU OpenMP on the Tilera TILEPro64. GPRM shows superior performance in almost all cases, only by controlling the number of tasks. GPRM also provides a low-overhead mechanism, called “Global Sharing”, which improves performance in multiprogramming situations. We use OpenMP, as the most popular model for shared-memory parallel programming as the main GPRM competitor for solving three well-known problems on both platforms: LU factorisation of Sparse Matrices, Image Convolution, and Linked List Processing. We focus on proposing solutions that best fit into the GPRM’s model of execution. GPRM outperforms OpenMP in all cases on the TILEPro64. On the Xeon Phi, our solution for the LU Factorisation results in notable performance improvement for sparse matrices with large numbers of small blocks. We investigate the overhead of GPRM’s task creation and distribution for very short computations using the Image Convolution benchmark. We show that this overhead can be mitigated by combining smaller tasks into larger ones. As a result, GPRM can outperform OpenMP for convolving large 2D matrices on the Xeon Phi. Finally, we demonstrate that our parallel worksharing construct provides an efficient solution for Linked List processing and performs better than OpenMP implementations on the Xeon Phi. The results are very promising, as they verify that our parallel programming framework for manycore processors is flexible and scalable, and can provide high performance without sacrificing productivity.

An investigation of the possible health-promoting modes of action of regular- and super- doses of phytase in the broiler chicken

The overall objective of this thesis was to study the effects of regular and high (super-) doses of phytase in the gut of broilers, with the aim of documenting the mechanism of their action leading to improvements in animal health. Phytase is often supplemented to commercial broiler diets to facilitate the hydrolysis of plant phytate and release of phosphorus for utilisation. Although not the original intention of its addition, phytase supplementation leads to improvements in growth performance parameters and enhanced nutrient utilisation. Further benefits have also been observed following the addition of super-doses of phytase which are not explained by an increase in phosphorus release, and thus have been termed ‘extra-phosphoric effects’. Using diets formulated to be adequate or marginally deficient in available phosphorus (aP; forming the negative control, NC), phytase was supplemented at 1,500 and 3,000 FTU/kg phytase in the first study (both super-doses) and the partitioning of nutrients within the body was investigated. It appeared that there were some metabolic changes between 1,500 and 3,000 FTU/kg, switching between protein and fat accretion, potentially as a consequence of nutrient availability, although these changes were not reflected by changes in growth performance parameters. However, the loss of the NC treatment without phytase on day 12 limits the comparison of the phytase within the NC treatment, but does allow for comparison of each dose at adequate or low dietary aP levels. As expected, a greater degree of phytate hydrolysis was achieved with 3,000 than with 1,500 FTU/kg phytase, but changes in carcass accretion characteristics were greater with 1,500 than 3,000 FTU/kg. Using these findings and the observation that there were no further changes in the parameters measured by increasing phytase from 1,500 to 3,000 FTU/kg (aside from phytate hydrolysis), 1,500 FTU/kg phytase was selected as the super-dose to be used in subsequent studies. The next study considered the influence of regular (500 FTU/kg) and super doses (1,500 FTU/kg) of phytase from within the gut. Overall, it was observed that changes were occurring to the gut environment, which ultimately would influence the absorptive capacity and conditions for further phytate hydrolysis. Dietary treatment influenced gut conditions such as pH, intestinal morphology and bacterial populations which can subsequently influence nutrient utilisation and potential for growth. The subsequent study was designed to investigate the effects within the gut in more detail. The release of nutrients from phytate hydrolysis and their bioavailability within the digesta can influence conditions within intestine, facilitating enhanced absorption. One of the parameters investigated was the expression of genes involved in the transport of nutrients in the intestine. Overall, there were few significant dietary treatment influences on gene expression in the intestine, however there was a dose-dependent response of phytase on the expression of the jejunual divalent mineral transporter. This indicates a change in divalent mineral bioavailability in the intestine, with correlations with inositol phosphate esters (IPs) being identified. This is likely explained by the IPs produced by phytase hydrolysis and accumulating in the digesta, differing between regular and high doses of phytase. It became apparent that interactions between the products of phytate hydrolysis (IP3, IP4) and minerals in the digesta had the potential to influence the gut environment and subsequent nutrient bioavailability and overall phytase action. The final study was designed to increase the content of the IPs, and investigate the influence of phytase under these conditions. As the complete hydrolysis of phytate to myo-inositol has been reported to be beneficial due to its proposed insulin mimetic effects, myo-inositol was also supplemented to one of the diets to see if any further benefits would be observed when supplemented alongside super-doses of phytase. Neither increased concentrations of the higher IP esters (IP6, IP5 and IP4) nor myo-inositol (myo-) had any effect on broiler growth performance, however there were still apparent beneficial influences of phytase supplementation. The results suggest considerable and important interactions between minerals and IP esters within the digesta, which ultimately have the potential to influence gut conditions and thus nutrient utilisation and growth performance. Reduced concentrations of blood glucose in the high IP ester diet with additional phytase supplementation suggest some insulin-like effects of myo- production. Additionally, the lack of effect of myo- supplementation on blood glucose and insulin concentrations suggests a difference between the structure of phytase-produced myo- and supplemented myo-. Although there were no improvements in growth performance by increasing phytase from 500 to 1,500 FTU/kg, there were changes occurring at the level of the gut and expression of genes in the intestine, influencing nutrient utilisation and the partitioning of nutrients within the body. There are many factors to be considered when supplementing phytase, with dietary nutrient content and nutrient release and IP production during phytate hydrolysis having an influence on phytase action, nutrient absorption and conditions within the gut. Super-doses of phytase may be beneficial for maintaining optimal gut conditions, clearing IP esters from the digesta, reducing their potential to form complexes with minerals and other nutrients, ultimately influencing the efficiency of production.