<p>Fast addition is essential in many DSP algorithms. Various structures have been introduced to speed up the time critical carry propagation. For high throughput applications, however, it may be necessary to introduce pipelining. In this report the power consumption of four different adder structures, with varying word length and different number of pipeline cuts, is compared. </p><p>Out of the four adder structures compared, the Kogge-Stone parallel prefix adder proves to be the best choice most of the time. The Brent-Kung parallel prefix adder is also a good choice, but the maximal throughput does not reach as high as the maximal throughput of the Kogge-Stone parallel prefix adder.</p>

Fast addition is essential in many DSP algorithms. Various structures have been introduced to speed up the time critical carry propagation. For high throughput applications, however, it may be necessary to introduce pipelining. In this report the power consumption of four different adder structures, with varying word length and different number of pipeline cuts, is compared. Out of the four adder structures compared, the Kogge-Stone parallel prefix adder proves to be the best choice most of the time. The Brent-Kung parallel prefix adder is also a good choice, but the maximal throughput does not reach as high as the maximal throughput of the Kogge-Stone parallel prefix adder.

Gegenstand der vorliegenden Arbeit ist die Präfigierung der Verben im Deutschen und Litauischen. Das Hauptgewicht dieser Arbeit liegt auf dem deutschen Präfix er- und seinen semantischen und syntaktischen Funktionen, die es zusammen mit einem Basisverb ausdrücken kann. Dann werden die Entsprechungen des Präfixes er- in der litauischen Sprache festgestellt und die Funktionen des Präfixes er- mit den Funktionen der entsprechenden litauischen Präfixen verglichen. / Priešdėlinių er- veiksmažodžių semantinės ir sintaksinės funkcijos vokiečių kalboje ir jų atitikmenys lietuvių kalboje. / Semantical and syntactical functions of the verbs with the prefix "er" in German and Lithuanian languages.

O prefixo ver- já foi tema de estudo para diversos linguistas que se dedicam à língua alemã (LEOPOLD, 1907; SCHIRMEIER, 2006, entre outros). O caráter polissêmico desse prefixo é o grande motivo para tamanho interesse. Diferentemente dos trabalhos citados, esta dissertação analisa novas formações verbais com o prefixo ver-, a fim de descobrir regularidades, tanto semântica quanto morfológicas no processo de formação dos verbos. Para tanto, pesquisamos novos verbos ainda não lexicalizados em um sítio eletrônico (http://www.deflok.de/HipHopLyrics/Texte/), composto por letras de música do hip hop alemão. Assim, o endereço virtual foi tomado como corpus para a tarefa. A busca permitiu encontrar doze novos verbos: verbassen, verspeien, verpeilen, verchecken, verächten, verkreuzen, verstumpfen, vergaukeln, verschlendern, vernabeln, verreimen, verstottern. Para tanto, os dicionários DUDEN (www.duden.de), Langescheidt (versão digital) e PONS (versão digital) foram usados como ferramenta de exclusão lexicográfica. O dicionário DUDEN funcionou também como fonte para conhecer as acepções que o prefixo ver- imprimia nos derivados em que se encontrava. O arcabouço teórico usado teve como base os conceitos da teoria da gramática de construções, a partir de GOLDBERG (1995) que permitiram conhecer novas propostas de interpretação dos mecanismos envolvidos na formação dos verbos com o prefixo ver. Ademais, recorremos à teoria da mesclagem de FAUCONNIER & TURNER (2003) para explicar a construção de sentido de determinados verbos, uma vez que essa permitia explorar aspectos não contemplados pela teoria da gramática de construções. / The prefix ver- has already been a theme of many linguistics (LEOPOLD, 1907; SCHIRMEIER, 2006, and others), that are dedicated to German language. The polissemic aspect from this prefix is the main reason for such a big interest. Unlike those authors, with this paper I analyse new verbal formations with the prefix ver- in order to discover semantic as well as morphological regularities in the process of verbal formations. Therefore, I researched new not yet lexicalized verbs in a website (http://www.deflok.de/HipHopLyrics/Texte/), composed of hip hop lyrics in German. The web address was treated as the corpus. Twelve new verbs were detected: verbassen, verspeien, verpeilen, verchecken, verächten, verkreuzen, verstumpfen, vergaukeln, verschlendern, vernabeln, verreimen, verstottern. In order to verify that these verbs were not yet lexicalized we used the following dictionaries: DUDEN (www.duden.de) and the digital version of Langenscheidt, and PONS. The dictionary DUDEN also served to specify the different meaning, that the prefix ver- gives to the derivates, in which it is part of. The theoretical support is based on the terms and definitions of the construction grammar (Goldberg, 1995), that allowed to get a new point of view about the mechanisms involved in the verb formation with the prefix ver-. Complementary I employed the definitions proposed by FAUCONNIER & TURNER (2003) to explain the meaning of certain neologism to which the construction grammar theory could not be applied.

The present thesis deals with the semantics of prefixes in Czech verbal neologisms. The first part describes verbal prefixation, the classification of prefixes, and their principal functions. Subsequently, polysemy, collocability and valency of prefixed verbs are characterized. The following section discusses the methodology, and characterizes the electronic neological database Neomat, which is the primary source of material for the thesis. The subject of analysis is neological verbs with prefixes od-, pro- and vy-. The meanings of these prefixes are described in relation to the collocability, valency and polysemy of prefixed verbs. The analysis aims to identify the principal meanings of the prefixes and to determine whether there are any meanings which have not been presented in any paper yet. Key words: neologism, nonce word, prefix, semantics, verb, word formation

Le stockage des données est un point crucial du développement des applications, et plus particulièrement des applications réparties. Les problématiques liées au stockage sont multiples : assurer la pérennité des données, en assurer l'identification et l'indexation, en garantir la recherche et l'accès, et le cas échéant le rappatriement afin d'être exploité par les différents acteurs de l'application. Il faut de surcroit concevoir des méthodes permettant de garantir l'efficacité de toutes des propriétés et opérations, dans un environnement multi-hétérogène, tant au niveau des formats des données, des protocoles d'échange, des systèmes et des applications. Dans cette thèse, nous proposons GRAPP&S, une architecture multi-échelle pour le stockage et l'indexation unifiée de différents formats de données tels que les fichiers, les flux de données, les requêtes sur les bases de données mais aussi l'accès à des services distants (des web services sur un serveur ou le cloud, ou des tâches de calcul dans un cluster HPC). GRAPP&S offre et orchestre une architecture hiérarchique de routage pour accéder aux différents types de données, et permet l'interconnexion de différentes communautés (réseaux locaux) grâce aux principes des systèmes multi-échelle. Les données sont présentées de façon transparente à l'utilisateur par l'intermédiaire de proxies spécifiques à chaque type de donnée. Le routage hiérarchique de GRAPP&S permet de s'affranchir de tables de routage, et d'éviter l'utilisation de solutions d'inondation pour la recherche de données.Enfin, nous proposons d'exploiter GRAPP&S dans le cadre particulier du E-Learning. Notre solution permettra de fédérer, à moindre coût, des ressources pédagogiques réparties sur plusieurs organisme, et d'en assurer l'exploitation par des apprenants. / Data storage is a crucial point for applications development and particularly for distributed applications. There are many issues related to data storage: the insurance of sustainability, identification and indexing of the data, the warranty of searching and accessing the data, and possibly the fetching of the data for applications use. So, there is a need to design methods for ensuring effectiveness of all of these properties and operations in heterogeneous environments, both in terms of data formats, exchange protocols, and applications.In this thesis, we propose GRAPP&S (Grid APPlication & Services), a multi-scale framework for an unified storage and indexing of both data and services. Indeed, GRAPP&S was designed to support different data formats such as files, stream or database queries, but also to support the access to distant services (web services, cloud services or HPC computing services, for example). GRAPP&S coordinates an hierarchical routing architecture that allows data indexing and access, thanks to a multi-scale network of local-area communities. Transparent access is provided by a network of specialized proxies, which handle most aspects related to data location, request handling, data preprocessing or summary and also data consistence.Finally, we exploit GRAPP&S in the particular context of E-Learning. Our solution will reduce the cost of merging distributed educational resources over several organizations and ensure the exploitation of learners..

This thesis presents a technique for automatic vectorization of innermost single statement loops with a cross-iteration data dependence by analyzing data-flow to recognize frequently recurring program idioms. Recognition is carried out by matching the circular SSA data-flow found around the loop body’s φ-function against several primitive patterns, forming a tree representation of the relevant data-flow that is then pruned down to a single parameterized node, providing a high-level specification of the data-flow idiom at hand used to guide algorithmic replacement applied to the intermediate representation. The versatility of the technique is shown by presenting an implementation supporting vectorization of both a limited class of linear recurrences as well as prefix sums, where the latter shows how the technique generalizes to intermediate representations with memory state in SSA-form. Finally, a thorough performance evaluation is presented, showing the effectiveness of the vectorization technique.

Testování multikomponentních IT a IoT systémů, které zpracovávají proudy různých zpráv je složitou úlohou. Proč je to složité? Kvůli množství komponentů, jejích asynchronní interakcí, různým kombinacím události pro testování, testovacímu prostředí, které se liší od reálného a množství dalším důvodů. Táhle práce navrhuje způsob generování komplexních dat pro testovací účely s minimálním zásahem vývojářů. Generování dat založeno na analýze sledu komunikace reálného systému a následnou syntézou podobných sledů pro testování. Práce také navrhuje framework na prvotní analýzu zpráv přenášených v zachycené komunikace. Tohle může být uděláno použitím různých abstraktních modelů: modelu zprávy a modelu komunikaci. Výsledkem téhle práce je implementovaní knihovna na tvorbu modelu zprávy a množina operací pro práci s tímto modelem.

Longest prefix matching is a well-studied problem in the context of IP-packet forwarding, an area of computational specialization where high performance with low memory impact is of the essence. Numerous specialized data structures exist for longest prefix matching in this setting, among them is Waldvogels binary search on lengths, WBSL for short. This data structure has previously been augmented with a Bloom filter to improve its performance, resulting in a new data structure: WBSL-BF. This study concerns the potential performance improvement of adding length-awareness to WBSL-BF. Performance of the augmented data structure is evaluated by recording the number of hash table accesses performed to carry out a series of queries, along with false positive rates for the Bloom filters. A test program is created to run multiple series of queries with different test configurations. The data structure is evaluated using two datasets containing prefixes from routing information base snapshots from real route collectors. The two datasets contain roughly 80000 and 900000 prefixes respectively. The results indicate that on the given datasets, augmenting the Bloom filter with length-awareness can reduce the number of hash table accesses when performing longest prefix matching by up to 44% in a best case scenario without increasing memory usage. However, the performance gain is limited in other scenarios, since the optimally configured standard Bloom filter leaves little to improve upon when applied to Waldvogels binary search on lengths.

In this study an attempt is made to describe the morphological apects of Tshiguvhu. In chapter 1, it is indicated that historically, there was extensive early contact between Vhaguvhu and Balobedu and Tlokwa. In ,chapters 2 and 3, nouns and pronouns are analysed morphologically. Some similarities and differences between Tshiguvhu and Tshivenda are highlighted. These differences are ascribed to influences from Lobedu and Tlokwa. In chapters 4 and 5, the form of the verb and the use of verb forms in various tenses, where applicable, are described morphologically. Some verb roots and extensions have been influenced by Northern Sotho dialects whereas some have not. In chapter 6, the morphology of adverbs, interrogatives, conjunctions, ideophones and interjections are briefly described. In conclusion, it is indicated that Tshiguvhu is a dialect of Venda because of its cultural and historical bonds with Venda, although linguistically it shares some features with certain Northern Sotho dialects. / African Languages / M.A. (African Languages)