Multiplication is one of the widely used arithmetic operations in a variety of applications, such as image/video processing and machine learning. FPGA vendors provide high-performance multipliers in the form of DSP blocks. These multipliers are not only limited in number and have fixed locations on FPGAs but can also create additional routing delays and may prove inefficient for smaller bit-width multiplications. Therefore, FPGA vendors additionally provide optimized soft IP cores for multiplication. However, in this work, we advocate that these soft multiplier IP cores for FPGAs still need better designs to provide high-performance and resource efficiency. Toward this, we present generic area-optimized, low-latency accurate, and approximate softcore multiplier architectures, which exploit the underlying architectural features of FPGAs, i.e., lookup table (LUT) structures and fast-carry chains to reduce the overall critical path delay (CPD) and resource utilization of multipliers. Compared to Xilinx multiplier LogiCORE IP, our proposed unsigned and signed accurate architecture provides up to 25% and 53% reduction in LUT utilization, respectively, for different sizes of multipliers. Moreover, with our unsigned approximate multiplier architectures, a reduction of up to 51% in the CPD can be achieved with an insignificant loss in output accuracy when compared with the LogiCORE IP. For illustration, we have deployed the proposed multiplier architecture in accelerators used in image and video applications, and evaluated them for area and performance gains. Our library of accurate and approximate multipliers is opensource and available online at https://cfaed.tu-dresden.de/pd-downloads to fuel further research and development in this area, facilitate reproducible research, and thereby enabling a new research direction for the FPGA community.

No description available.

Since their emergence almost two decades ago, accelerators have evolved towards a global phenomenon that has the potential to shape global economies and societies. The rapid evolvement of the accelerator phenomenon increasingly caught scholars’ interest to examine what makes this specific form of startup support unique and popular in practice. Despite the growing amount of valuable contributions in accelerator research, several key aspects of the accelerator phenomenon remain unsolved or misunderstood, which leaves scholars as well as practitioners with a discordant understanding of how and why accelerators evolved and how they function to date. Hence, ambiguities are still large with regard to thoroughly understanding the accelerator mechanism and its recent developments. However, this understanding is crucial to develop theory in accelerator research and provide assistance for practitioners on how to increase value creation of accelerator programs. By means of five research studies this cumulative dissertation contributes to the lack of common conceptual foundations, an adequate consideration of theoretical underpinnings, a contemporary presentation of defining accelerator characteristics and their evolvement, and reliable insights on accelerator efficacy. In addition, this dissertation contributes numerous lessons learned and guidance to various stakeholders affected by accelerators.

Ultra-high dose rate proton radiation has the potential to improve cancer treatment by reducing the normal tissue complication probability and, at the same time, reaching the tumor control probability known from conventional photon radiation therapy. Here, the ultra-high dose rate leads to normal tissue sparing via the FLASH effect. Before a clinical implementation is possible, the proton FLASH effect requires translational research via in-vivo irradiation studies with small animals. Laser plasma-based accelerators (LPAs) for protons offer unique opportunities for studying the proton FLASH effect, since the proton dose rate at LAPs is in the order of 10^9 Gy/s, which is unreached at conventional medical proton accelerators. Different to medical proton accelerators, LPAs are operated in a pulsed mode and feature a lower beam stability caused by inherent pulse-to-pulse fluctuations of the acceleration process. To ensure successful in-vivo irradiation studies, advanced beam delivery, monitoring and dosimetry concepts for an online-monitored application of the 3D dose distribution in the target volume (TV) of the in-vivo sample are needed. The detectors and dosimetric concept developed in this thesis enable the world wide frst pilot radiobiological in-vivo study with LPA protons, where mouse ear tumors are irradiated with ultra-high dose rate proton pulses. For performing the radiobiological study, the ALBUS-2S (Advanced Laser-driven Beamlines for User-specifc Studies - 2 Solenoids) beamline is used, which is installed at the compact petawatt (PW) laser system DRACO (Dresden laser acceleration source) at HZDR (Helmholtz-Zentrum Dresden-Rossendorf). In this thesis, a scintillator-based time-of-fight (ToF) beam monitoring sytem (BMS) is developed, which records single-pulse proton energy spectra in transmission at the ALBUS-2S beamline. A relative energy uncertainty of 5.5 % (1σ) is reached for the ToF BMS, allowing for a Monte Carlo simulation-based prediction of depth dose profiles at the irradiation site. The ToF BMS is used for characterization of the ALBUS-2S LPA beamline for application-oriented parameters, in order to qualify the LPA proton source for radiobiological in-vivo studies. Furthermore, a dosimetry and beam monitoring concept for in-vivo irradiations of small target volumes with LPA protons is presented in this thesis. With the overall relative dose uncertainty of 7.4 % (2σ) for the specifc mouse ear tumor irradiation scenario, the concept enables verifcation of accurate volumetric dose delivery to the mm-scale TVs. In addition, tomography-based approaches with scintillators are investigated as detectors for online 3D dose measurement at LPAs. The miniature scintillator dosimeter (miniSCIDOM) detector, which is developed in the scope of this thesis, is used for online 3D dose measurements in 1 cm^3 volumes with < 1 mm^3 resolution at the irradiation site of the ALBUS-2S beamline. For online 3D dose measurements directly behind the LPA proton source of the DRACO PW laser system, the optical cone beam tomograph for proton online dosimetry (OCTOPOD) detector is developed. The OCTOPOD detector has a sensitive volume of 5 cm-diameter and water equivalent thickness of 4.3 cm, which is sufficient to stop 70 MeV protons. It is designed to reach a spatial resolution of 1 mm^3. The detectors developed in this thesis are optimized tools for source-to-sample characterization of LPA beamlines and hence are an essential contribution for radiobiological in-vivo studies with LPA protons.

Ce travail se place dans le contexte du détecteur de vertex (VXD) composé de capteurs CMOS pour l'ILC, et dans celui du télescope en faisceau du projet européen AIDA. La thèse inclut les tests en faisceau des éléments du télescope AIDA : les super-plans SALAT et les échelles double faces PLUME. Elle questionne la valeur ajoutée en terme d'alignement, des couches double faces de capteurs CMOS pour le VXD de l'ILD. Une nouvelle méthode d'alignement autonome de chacune des 3 double couches du VXD grâce aux mini-vecteurs construits sur chaque zone de recouvrement inter-échelle est proposée et a été testée avec des particules de haute impulsion. Cependant, seules les particules du bruit de fond faisceau, de plus basses impulsions, permettent l'obtention d'une statistique suffisante pour cet alignement. Ce bruit de fond a alors été étudié et une estimation des taux d'occupation des capteurs du VXD a conduit à une ré-estimation des vitesses de lecture des capteurs de chaque couche du VXD. / This work was conducted in the context of a vertex detector (VXD) composed of CMOS sensors for ILD and in the context of the beam telescope of the european project AIDA. The provides the results of beam tests for the new telescope components : the SALAT super-planes and the PLUME double sided ladders. The thesis adresses the added value in terms of alignment, of double sided layers of CMOS sensors for the VXD of ILD. A new standalone alignment method of each of the three double sided layers of VXD with the mini-vectors built on each overlapping zone between the consecutive ladders is analysed. Such alignment was validated with high momentum particles. However, only the beam background particles, with lower momentum, can provide the minimum statistic for this kind of alignment. Thus, the beam background noise was studied and the occupancy rate of the VXD sensors was studied. This led to a reassessment of the readout speed for the sensors of each layer of the VXD.

HESS J1741−302 ist eine nicht identifizierte Quelle sehr hochenergetischer Gammastrahlen, welche circa 1,7 Grad vom Zentrum der Milchstraße entfernt liegt. Diese Quelle ist eines der schwächsten Objekte im TeV-Bereich mit einem Photonfluss von Φ(>1 TeV) = (1.65 ± 0.28stat ± 0.33sys) × 10^−13 cm^−2 s^−1, was ~1% des Krebsnebelflusses im gleichen Energiebereich entspricht. Die Analyse des aktuellen H.E.S.S. Datensatzes von 145 Stunden Beobachtungen mit hoher Qualität gibt Einblicke in die Morphologie von HESS J1741−302. Das Energiespektrum von HESS J1741−302 geht über 10 TeV hinaus, ohne dabei ein klares Anzeichen für einen spektralen Abbruch zu zeigen. Das Spektrum kann durch ein Potenzgesetz mit einem spektralen Index von Γ = 2.28 ± 0.16stat ± 0.20sys und einer Normierung bei 1 TeV von Φ0 = (2.12 ± 0.42stat ± 0.42sys) × 10^−13 cm^−2 s^−1 TeV^−1 beschrieben werden. In der vorliegenden Arbeit werden verschiedene Szenarien für die beobachtete Gammastrahlung und deren Entstehung in Betracht gezogen. Diese beinhalten die Wechselwirkung von Protonen der kosmischen Strahlung mit Molekülwolken entlang der Sichtlinie, IC Streuung an Infrarot-Photonen eines nahe gelegenen OH/IR Sterns und die Präsenz eines Pulsarwindnebels, welcher möglicherweise zu PSR B1737−30 gehört. / HESS J1741−302 is an unidentified very-high-energy (VHE) γ-ray source located in the Galactic Plane at about 1.7° away from the Galactic Center. It is one of the faintest TeV objects detected so far, with a flux Φ(>1 TeV) = (1.65 ± 0.28stat ± 0.33sys) × 10^−13 cm^−2 s^−1 corresponding to ~ 1% of the Crab Nebula flux at the same energies. The data analysis of an updated high-quality dataset of ~145 hours of VHE H.E.S.S. data taken between 2004 and 2013 has revealed the morphology of HESS J1741−302. The γ-ray spectrum of HESS J1741−302 extends beyond 10 TeV without showing any clear evidence of a cut-off. The source spectrum is well described by a power-law model with a spectral index of Γ = 2.28 ± 0.16stat ± 0.20sys and a normalization at 1 TeV of Φ0 = (2.12 ± 0.42stat ± 0.42sys) × 10^−13 cm^−2 s^−1 TeV^−1. Different scenarios will be considered in this thesis, including the interaction of cosmic-ray protons with molecular clouds found along the line of sight, inverse Compton scattering of infra-red photons provided by a nearby OH/IR star and the presence of a nearby pulsar wind nebula possibly related to PSR B1737−30, in order to explain the observed VHE gamma-ray emission.

<p>Metalloproteins, or proteins with a metal ion cofactor, are essential for biological function of both lower and higher level organisms. These proteins provide a multitude of functions from molecular transport, such as the hemoglobin transport of oxygen, to biologically important catalytic processes. As an example case, photosystem II (PSII) is studied as a representative metalloprotein. It was chosen based on the potential impact in the energy sector due to its ability to perform water oxidation using solar based energy. Understanding mechanisms by which the Mn₄Ca cluster inside PSII, also known as the oxygen evolving complex (OEC), can store energy as redox equivalents for splitting water will be essential for future development of analogous artificial systems. By using time resolved x-ray spectroscopy, the electron structure of the metal in the protein was probed through the catalytic cycle. While the applications mentioned herein are based on PSII from spinach, the developments in time-resolved x-ray spectroscopy techniques are also applicable to other metalloproteins.</p><p></p><p>By creating a new x-ray spectrometer we were able to capture the difference in x-ray emission spectra between two compounds differing in a single metal bound ligand, i.e. Mn^IV-OH and Mn^IV=O. This both establishes the functionality of the x-ray emission spectrometer and provides useful insight into the expected changes upon an oxygen double bond formation. This change in spectroscopic signal is discussed in context of the OEC which has been hypothesized to form a Mn^IV=O state.</p><p></p><p>A new sample delivery system and further developments to the x-ray spectrometer enabled both time-resolved x-ray absorption and time-resolved x-ray emission of PSII. These experiments show the potential of synchrotron sources for time-resolved x-ray spectroscopy. From our x-ray absorption measurements we were able to follow the electronic structure changes in time using a single incident photon energy. From the kinetic traces obtained, we show possible alternative interpretations of previous results showing a delay in reduction during the final step in water oxidation. From the x-ray emission spectroscopy (XES) measurements of PSII we were able to reproduce previous results within a limited collection time and give estimates for data size requirements for metalloproteins using this spectrometer. Between the results of both these measurements, we show the improved capability for time resolved measurements at synchrotrons.</p><p>The development of x-ray free electron lasers (XFELs) has also opened many opportunities for understanding faster electronic dynamics by providing femtosecond x-ray pulse durations with ~10¹² photons per pulse. While theoretical modeling of distortions to crystallographic data have been performed, little to no work has been done to understand under what conditions such an intense pulse will have on an impact on emission spectra. Here an atomistic model was developed, and data collected, to clarify the effects of sequential ionization, i.e. two single photons absorbed by the same atom at different times during a single pulse. Experimentally we found that XFELs easily achieve flux densities that invoke a different response than is classically observed for single photon absorption and emission for Mn^II which was used as a representative case for 3d transition metals in general. We also give parameters by which the onset of this damage can be predicted and an approximation to its effect on 3d transition metals. Additionally this work guides the work of future XFEL facilities as it shows that shorter pulses, currently believed to be able to escape x-ray induced distortions to crystallography data, is not a viable method for overcoming changes in x-ray emission spectra.</p><div><br></div>

L'accélération matérielle désigne l'utilisation d'architectures spécialisées pour effectuer certaines tâches plus vite ou plus efficacement que sur du matériel générique. Les accélérateurs ont traditionnellement été utilisés dans des environnements contraints en ressources, comme les systèmes embarqués. Cependant, avec la fin des règles empiriques ayant régi la conception de matériel pendant des décennies, ces quinze dernières années ont vu leur apparition dans les centres de calcul et des environnements de calcul haute performance. Les FPGAs constituent une plateforme d'implémentation commode pour de tels accélérateurs, autorisant des compromis subtils entre débit/latence, surface, énergie, précision, etc. Cependant, identifier de bons compromis représente un défi, dans la mesure où l'espace de recherche est généralement très large. Cette thèse propose des techniques de conception pour résoudre ce problème. Premièrement, nous nous intéressons aux compromis entre performance et précision pour la conversion flottant vers fixe. L'utilisation de l'arithmétique en virgule fixe au lieu de l'arithmétique flottante est un moyen efficace de réduire l'utilisation de ressources matérielles, mais affecte la précision des résultats. La validité d'une implémentation en virgule fixe peut être évaluée avec des simulations, ou en dérivant des modèles de précision analytiques de l'algorithme traité. Comparées aux approches simulatoires, les méthodes analytiques permettent une exploration plus exhaustive de l'espace de recherche, autorisant ainsi l'identification de solutions potentiellement meilleures. Malheureusement, elles ne sont applicables qu'à un jeu limité d'algorithmes. Dans la première moitié de cette thèse, nous étendons ces techniques à des filtres linéaires multi-dimensionnels, comme des algorithmes de traitement d'image. Notre méthode est implémentée comme une analyse statique basée sur des techniques de compilation polyédrique. Elle est validée en la comparant à des simulations sur des données réelles. Dans la seconde partie de cette thèse, on se concentre sur les stencils itératifs. Les stencils forment un motif de calcul émergeant naturellement dans de nombreux algorithmes utilisés en calcul scientifique ou dans l'embarqué. À cause de cette diversité, il n'existe pas de meilleure architecture pour les stencils de façon générale : chaque algorithme possède des caractéristiques uniques (intensité des calculs, nombre de dépendances) et chaque application possède des contraintes de performance spécifiques. Pour surmonter ces difficultés, nous proposons une famille d'architectures pour stencils. Nous offrons des paramètres de conception soigneusement choisis ainsi que des modèles analytiques simples pour guider l'exploration. Notre architecture est implémentée sous la forme d'un flot de génération de code HLS, et ses performances sont mesurées sur la carte. Comme les résultats le démontrent, nos modèles permettent d'identifier les solutions les plus intéressantes pour chaque cas d'utilisation. / Hardware acceleration is the use of custom hardware architectures to perform some computations faster or more efficiently than on general-purpose hardware. Accelerators have traditionally been used mostly in resource-constrained environments, such as embedded systems, where resource-efficiency was paramount. Over the last fifteen years, with the end of empirical scaling laws, they also made their way to datacenters and High-Performance Computing environments. FPGAs constitute a convenient implementation platform for such accelerators, allowing subtle, application-specific trade-offs between all performance metrics (throughput/latency, area, energy, accuracy, etc.) However, identifying good trade-offs is a challenging task, as the design space is usually extremely large. This thesis proposes design methodologies to address this problem. First, we focus on performance-accuracy trade-offs in the context of floating-point to fixed-point conversion. Usage of fixed-point arithmetic instead of floating-point is an affective way to reduce hardware resource usage, but comes at a price in numerical accuracy. The validity of a fixed-point implementation can be assessed using either numerical simulations, or with analytical models derived from the algorithm. Compared to simulation-based methods, analytical approaches enable more exhaustive design space exploration and can thus increase the quality of the final architecture. However, their are currently only applicable to limited sets of algorithms. In the first part of this thesis, we extend such techniques to multi-dimensional linear filters, such as image processing kernels. Our technique is implemented as a source-level analysis using techniques from the polyhedral compilation toolset, and validated against simulations with real-world input. In the second part of this thesis, we focus on iterative stencil computations, a naturally-arising pattern found in many scientific and embedded applications. Because of this diversity, there is no single best architecture for stencils: each algorithm has unique computational features (update formula, dependences) and each application has different performance constraints/requirements. To address this problem, we propose a family of hardware accelerators for stencils, featuring carefully-chosen design knobs, along with simple performance models to drive the exploration. Our architecture is implemented as an HLS-optimized code generation flow, and performance is measured with actual execution on the board. We show that these models can be used to identify the most interesting design points for each use case.

19 Dec 2004. / Published through the Information Bridge: DOE Scientific and Technical Information. "IS-T 1936" Jongik Park. 12/19/2004. Report is also available in paper and microfiche from NTIS.

Thesis (Ph.D.); Submitted to Univ. of California, Berkeley, CA (US); 1 Jun 2003. / Published through the Information Bridge: DOE Scientific and Technical Information. "LBNL--54350" Scielzo, Nicholas David. USDOE Director. Office of Science. Nuclear Physics (US) 06/01/2003. Report is also available in paper and microfiche from NTIS.