Zugl.: Kassel, Univ., Diss., 2008

No description available.

No description available.

No description available.

Dans le domaine du calcul haute performance, de nombreux programmes étalons ou benchmarks sont utilisés pour mesurer l’efficacité des calculateurs,des compilateurs et des optimisations de performance. Les benchmarks de référence regroupent souvent des programmes de calcul issus de l’industrie et peuvent être très longs. Le processus d’´étalonnage d’une nouvelle architecture de calcul ou d’une optimisation est donc coûteux.La plupart des benchmarks sont constitués d’un ensemble de noyaux de calcul indépendants. Souvent l’´étalonneur n’est intéressé que par un sous ensemble de ces noyaux, il serait donc intéressant de pouvoir les exécuter séparément. Ainsi, il devient plus facile et rapide d’appliquer des optimisations locales sur les benchmarks. De plus, les benchmarks contiennent de nombreux noyaux de calcul redondants. Certaines opérations, bien que mesurées plusieurs fois, n’apportent pas d’informations supplémentaires sur le système à étudier. En détectant les similarités entre eux et en éliminant les noyaux redondants, on diminue le coût des benchmarks sans perte d’information.Cette thèse propose une méthode permettant de décomposer automatiquement une application en un ensemble de noyaux de performance, que nous appelons codelets. La méthode proposée permet de rejouer les codelets,de manière isolée, dans différentes conditions expérimentales pour pouvoir étalonner leur performance. Cette thèse étudie dans quelle mesure la décomposition en noyaux permet de diminuer le coût du processus de benchmarking et d’optimisation. Elle évalue aussi l’avantage d’optimisations locales par rapport à une approche globale.De nombreux travaux ont été réalisés afin d’améliorer le processus de benchmarking. Dans ce domaine, on remarquera l’utilisation de techniques d’apprentissage machine ou d’´echantillonnage. L’idée de décomposer les benchmarks en morceaux indépendants n’est pas nouvelle. Ce concept a été aappliqué avec succès sur les programmes séquentiels et nous le portons à maturité sur les programmes parallèles.Evaluer des nouvelles micro-architectures ou la scalabilité est 25× fois plus rapide avec des codelets que avec des exécutions d’applications. Les codelets prédisent le temps d’exécution avec une précision de 94% et permettent de trouver des optimisations locales jusqu’`a 1.06× fois plus efficaces que la meilleure approche globale. / In high performance computing, benchmarks evaluate architectures, compilers and optimizations. Standard benchmarks are mostly issued from the industrial world and may have a very long execution time. So, evaluating a new architecture or an optimization is costly. Most of the benchmarks are composed of independent kernels. Usually, users are only interested by a small subset of these kernels. To get faster and easier local optimizations, we should find ways to extract kernels as standalone executables. Also, benchmarks have redundant computational kernels. Some calculations do not bring new informations about the system that we want to study, despite that we measure them many times. By detecting similar operations and removing redundant kernels, we can reduce the benchmarking cost without loosing information about the system. This thesis proposes a method to automatically decompose applications into small kernels called codelets. Each codelet is a standalone executable that can be replayed in different execution contexts to evaluate them. This thesis quantifies how much the decomposition method accelerates optimization and benchmarking processes. It also quantify the benefits of local optimizations over global optimizations. There are many related works which aim to enhance the benchmarking process. In particular, we note machine learning approaches and sampling techniques. Decomposing applications into independent pieces is not a new idea. It has been successfully applied on sequential codes. In this thesis we extend it to parallel programs. Evaluating scalability or new micro-architectures is 25× faster with codelets than with full application executions. Codelets predict the execution time with an accuracy of 94% and find local optimizations that outperform the best global optimization up to 1.06×.

O tratamento biológico anaeróbio aparece como uma opção importante para remover a carga orgânica poluidora da vinhaça, e vem despertando interesse entre pesquisadores do mundo todo devido aos sucessos já alcançados, principalmente em trabalhos científicos. O interesse científico aponta para o aprimoramento do processo de degradação desse efluente em taxas elevadas, além de promover a produção de biogás, com potencial para produção de energia. A tecnologia de tratamento anaeróbio mais utilizada para o tratamento da vinhaça é o reator UASB, apresentando resultados satisfatórios inclusive em unidades em escala plena. A interrupção do funcionamento do reator em períodos de entressafra é, indubitavelmente, um problema a ser enfrentado na operação do reator UASB, cujo período de partida é lento. A repartida do sistema é agravada nos reatores de alta taxa, requerendo estratégias adequadas para reduzir o período necessário permitindo mitigar prontamente os efeitos do aporte de vinhaça ao ambiente e recuperar energia a partir dessa matriz. Diante disso, um reator anaeróbio de fluxo ascendente com manta de lodo (UASB) em escala piloto foi utilizado para o estudo da repartida do sistema e do tratamento da vinhaça e do melaço, sendo o último utilizado como fonte de carbono no período relativo à entressafra. O reator UASB foi operado durante 335 dias. Na Fase I utilizou-se a vinhaça como afluente, na Fase II o melaço de cana-de-açúcar e na Fase III novamente a vinhaça. Após período de interrupção de 30 dias, o período de repartida do sistema em estudo foi de 69 dias. Logo no início da repartida do reator, a eficiência de remoção de DQO foi superior a 80%, alcançando eficiência de 89,95 % em apenas oito dias de operação. O reator UASB apresentou valor médio de eficiência de remoção de DQO para as Fases I (vinhaça), II (melaço) e III (vinhaça) de 84 ± 3,35%; 82 ± 5,12% e 80 ± 9,22%, para cargas aplicadas de 5 a 34 kgDQO.m-3.d-1, 33 kgDQO.m-3.d-1 e 5 a 19 kgDQO.m-3.d-1, respectivamente. Em relação à produção de metano, de todo o período operacional, a Fase I referente à repartida do reator utilizando vinhaça como substrato após um mês sem alimentação foi a que apresentou a maior produção média no período, 7,93 LCH4. (L.d) -1, devido aos maiores valores de carga orgânica volumétrica (COV) aplicadas (5 a 34 kgDQO.m-3.d-1). A Fase II referente ao período de entressafra utilizando o melaço, foi de 7,28 LCH4. (L.d) -1 devido ao acúmulo de ácidos voláteis totais. A Fase III com retorno da safra utilizando a vinhaça, foi de 1,23 LCH4. (L.d) -1 devido aos menores valores de carga aplicada ao reator (5 a 8,5 kgDQO.m-3.d-1). As análises microbiológicas mostraram elevada diversidade microbiana e permitiu observar que com a utilização do melaço ocorreu alteração da população de microrganismos presente no reator, que apresentaram longo período de adaptação quando submetidos novamente à vinhaça, na última fase de operação. De forma geral, os resultados apontam que a tecnologia UASB é adequada no tratamento de vinhaça e melaço de alta carga orgânica e na geração de biogás. A substituição por melaço permitiu obter produção de biogás semelhante ao uso da vinhaça, entretanto dificultou a repartida, quando substituído pela vinhaça novamente (Fase III). / Biological anaerobic treatment systems are a promising approach for the removal of organic loading content of vinasse. Previous studies have demonstrated the optimization of the vinasse degradation process at high rates, promoting the biogas production with potential for energy use. Upflow anaerobic with sludge blanket (UASB) is the most applied anaerobic technology for the treatment of vinasse. Such reactor configuration presents satisfactory results even in full-scale units. The interruption of reactor operation during the off-season is undoubtedly an operational problem to be faced, whose restarting period is slow. The system restarting is intensified in high-rate reactors, requiring adequate strategies to reduce the time needed to mitigate the environmental effects of vinasse supply and to recover energy from that matrix. Therefore, a pilot-scale UASB was used to investigate the system restart and the treatment of vinasse and molasses. Sugarcane molasses was used as the carbon source during the off-season period. The system was operated during 335 days. The operational procedure consisted of three phases, in which different streams were used as carbon source: vinasse (Phases I), sugarcane molasses (Phases II) and vinasse (Phases III). After an interruption period of 30 days, the restarting period was 69 days. At the beginning of the restarting period, the average COD removal efficiency was higher than 80%, reaching COD removal efficiencies of 89.95% in only eight days of operation. The average COD removal efficiencies for Phases I (vinasse), II (molasses) and III (vinasse) were 84 ± 3.35%; 82 ± 5,12% and 80 ± 9,22%, respectively. Those removal efficiencies were achieved with organic loading rate 5 to 34 kgCOD.m-3.d-1 , 33 kgCOD.m-3.d-1 and 5 to 19 kgCOD.m-3.d-1, respectively. Regarding methane (CH4) production, Phase I (vinasse as substrate after one month without feeding) presented the highest average CH4 production (7.93 LCH4. (L.d) -1), due to the higher values of OLR (5 to 34 kgCOD.m-3.d-1). The CH4 production during Phase II (off-season using molasses as substrate) was 7.28 LCH4. (L.d) -1, due to the accumulation of volatile fatty acids (VFA). In Phase III, in which vinasse was applied again as substrate the CH4 production was 1.23 LCH4.(L.d)-1, due to the lower values of organic loading rate to the reactor (5 to 8.5 kgCOD.m-3.d-1). Microbiological analyses showed high microbial diversity in biomass developed at UASB reactor. Moreover, it was possible to observe that with molasses used as carbon source probably promotes changes in the microorganisms population, After period with molasses feeding, such microbiota presented a long adaptation period when resubmitted to vinasse, as the main carbon source in the feeding. In general, the results indicate that the UASB configuration was adequate for the vinasse and molasses treatment at high organic loading rate, resulting in high biogas production. In which the feeding with sugarcane molasses resulted in a biogas production similar to that obtained by vinasse feeding, but it made it difficult to divide when replaced by vinasse (Phase III).

Kernel patches are frequently released to fix security vulnerabilities and bugs. However, users and system administrators often delay installing these updates because they require a system reboot, which results in disruption of service and the loss of application state. Unfortunately, the longer an out-of-date system remains operational, the higher is the likelihood of a system being exploited. Approaches, such as dynamic patching and hot swapping, have been proposed for updating the kernel. All of them either limit the types of updates that are supported, or require significant programming effort to manage. We have designed a system that checkpoints application-visible state, updates the kernel, and restores the application state. By checkpointing high-level state, our system no longer depends on the precise implementation of a patch and can apply all backward compatible patches. The results show that updates to major kernel releases can be applied with minimal changes.

The runtime for solving constraint satisfaction problems (CSP) and propositional satisfiability problems (SAT) using systematic backtracking search has been shown to exhibit great variability. Randomization and restarts is an effective technique for reducing such variability to achieve better expected performance. Several restart strategies have been proposed and studied in previous work and show differing degrees of empirical effectiveness. <br /><br /> The first topic in this thesis is the extension of analytical results on restart strategies through the introduction of physically based assumptions. In particular, we study the performance of two of the restart strategies on Pareto runtime distributions. We show that the geometric strategy provably removes heavy tail. We also examine several factors that arise during implementation and their effects on existing restart strategies. <br /><br /> The second topic concerns the development of a new hybrid restart strategy in a realistic problem setting. Our work adapts the existing general approach on dynamic strategy but implements more sophisticated machine learning techniques. The resulting hybrid strategy shows superior performance compared to existing static strategies and an improved robustness.

The runtime for solving constraint satisfaction problems (CSP) and propositional satisfiability problems (SAT) using systematic backtracking search has been shown to exhibit great variability. Randomization and restarts is an effective technique for reducing such variability to achieve better expected performance. Several restart strategies have been proposed and studied in previous work and show differing degrees of empirical effectiveness. <br /><br /> The first topic in this thesis is the extension of analytical results on restart strategies through the introduction of physically based assumptions. In particular, we study the performance of two of the restart strategies on Pareto runtime distributions. We show that the geometric strategy provably removes heavy tail. We also examine several factors that arise during implementation and their effects on existing restart strategies. <br /><br /> The second topic concerns the development of a new hybrid restart strategy in a realistic problem setting. Our work adapts the existing general approach on dynamic strategy but implements more sophisticated machine learning techniques. The resulting hybrid strategy shows superior performance compared to existing static strategies and an improved robustness.

University, Diss--Kassel, 2008.