Techn. University, Diss., 2007--Darmstadt.

This thesis presents methods for automated synthesis of flexible chip multiprocessor systems from parallel programs targeted at FPGAs to exploit both task-level parallelism and architecture customization. Automated synthesis is necessitated by the complexity of the design space. A detailed description of the design space is provided in order to determine which parameters should be modeled to facilitate automated synthesis by optimizing a cost function, the emphasis being placed on inclusive modeling of parameters from application, architectural and physical subspaces, as well as their joint coverage in order to avoid pre-constraining the design space. Given a parallel program and a set of an IP library, the automated synthesis problem is to simultaneously (i) select processors (ii) map and schedule tasks to them, and (iii) select one or several networks for inter-task communications such that design constraints and optimization objectives are met. The research objective in this thesis is to find a suitable model for automated synthesis, and to evaluate methods of using the model for architectural optimizations. Our contributions are a holistic approach for the design of such systems, corresponding models to facilitate automated synthesis, evaluation of optimization methods using state of the art integer linear and answer set programming, as well as the development of synthesis heuristics to solve runtime challenges. / Aktuelle Technologien erlauben es komplexe Multiprozessorsysteme auf einem Chip mit Milliarden von Transistoren zu realisieren. Der Entwurf solcher Systeme ist jedoch zeitaufwendig und schwierig. Diese Arbeit befasst sich mit der Frage, wie On-Chip Multiprozessorsysteme ausgehend von parallelen Programmen automatisch synthetisiert werden können. Die Implementierung der Multiprozessorsysteme auf rekonfigurierbaren Chips erlaubt es die gesamte Architektur an die Struktur eines vorliegenden parallelen Programms anzupassen. Auf diese Weise ist es möglich die aktuellen technologischen Unzulänglichkeiten zu umgehen, insbesondere die nicht weitersteigende Taktfrequenzen sowie den langsamen Zugriff auf Datenspeicher. Eine Automatisierung des Entwurfs von Multiprozessorsystemen ist notwendig, da der Entwurfsraum von Multiprozessorsystemen zu groß ist, um vom Menschen überschaut zu werden. In einem ersten Ansatz wurde das Syntheseproblem mittels linearer Gleichungen modelliert, die dann durch lineare Programmierungswerkzeuge gelöst werden können. Ausgehend von diesem Ansatz wurde untersucht, wie die typischerweise langen Rechenzeiten solcher Optimierungsmethoden durch neuere Methode aus dem Gebiet der Erfüllbarkeitsprobleme der Aussagenlogik minimiert werden können. Dabei wurde die Werkzeugskette Potassco verwendet, in der lineare Programme direkt in Logikprogramme übersetzt werden können. Es wurde gezeigt, dass dieser zweite Ansatz die Optimierungszeit um bis zu drei Größenordnungen beschleunigt. Allerdings lassen sich große Syntheseprobleme auf diese weise wegen Speicherbegrenzungen nicht lösen. Ein weiterer Ansatz zur schnellen automatischen Synthese bietet die Verwendung von Heuristiken. Es wurden im Rahmen diese Arbeit drei Heuristiken entwickelt, die die Struktur des vorliegenden Syntheseproblems ausnutzen, um die Optimierungszeit zu minimieren. Diese Heuristiken wurden unter Berücksichtigung theoretischer Ergebnisse entwickelt, deren Ursprung in der mathematische Struktur des Syntheseproblems liegt. Dadurch lassen sich optimale Architekturen in kurzer Zeit ermitteln. Die durch diese Dissertation offen gewordene Forschungsarbeiten sind u. a. die Berücksichtigung der zeitlichen Reihenfolge des Datenaustauschs zwischen parallelen Tasks, die Optimierung des logik-basierten Ansatzes, die Integration von Prozessor- und Netzwerksimulatoren zur funktionalen Verifikation synthetisierter Architekturen, sowie die Entwicklung geeigneter Architekturkomponenten.

The ever-growing demand for more computing power forces hardware vendors to put an increasing number of multiprocessors into a single server system, which usually exhibits a non-uniform memory access (NUMA). In-memory database systems running on NUMA platforms face several issues such as the increased latency and the decreased bandwidth when accessing remote main memory. To cope with these NUMA-related issues, a DBMS has to allow flexible data partitioning and data placement at runtime. In this demonstration, we present ERIS, our NUMA-aware in-memory storage engine. ERIS uses an adaptive partitioning approach that exploits the topology of the underlying NUMA platform and significantly reduces NUMA-related issues. We demonstrate throughput numbers and hardware performance counter evaluations of ERIS and a NUMA-unaware index for different workloads and configurations. All experiments are conducted on a standard server system as well as on a system consisting of 64 multiprocessors, 512 cores, and 8 TBs main memory.

A current trend in the semiconductor industry is the use of Multi-Processor Systems-on-Chip (MPSoCs) for a wide variety of applications such as image processing, automotive, multimedia, and robotic systems. Most applications gain performance advantages by executing parallel tasks on multiple processors due to the inherent parallelism. Moreover, heterogeneous structures provide high performance/energy efficiency, since application-specific processing elements (PEs) can be exploited. The increasing number of heterogeneous PEs leads to challenging communication requirements. To overcome this challenge, Networks-on-Chip (NoCs) have emerged as scalable on-chip interconnect. Nevertheless, NoCs have to deal with many design parameters such as virtual channels, routing algorithms and buffering techniques to fulfill the system requirements. This thesis highly contributes to the state-of-the-art of FPGA-based MPSoCs and NoCs. In the following, the three major contributions are introduced. As a first major contribution, a novel router concept is presented that efficiently utilizes communication times by performing sequences of arithmetic operations on the data that is transferred. The internal input buffers of the routers are exchanged with processing units that are capable of executing operations. Two different architectures of such processing units are presented. The first architecture provides multiply and accumulate operations which are often used in signal processing applications. The second architecture introduced as Application-Specific Instruction Set Routers (ASIRs) contains a processing unit capable of executing any operation and hence, it is not limited to multiply and accumulate operations. An internal processing core located in ASIRs can be developed in C/C++ using high-level synthesis. The second major contribution comprises application and performance explorations of the novel router concept. Models that approximate the achievable speedup and the end-to-end latency of ASIRs are derived and discussed to show the benefits in terms of performance. Furthermore, two applications using an ASIR-based MPSoC are implemented and evaluated on a Xilinx Zynq SoC. The first application is an image processing algorithm consisting of a Sobel filter, an RGB-to-Grayscale conversion, and a threshold operation. The second application is a system that helps visually impaired people by navigating them through unknown indoor environments. A Light Detection and Ranging (LIDAR) sensor scans the environment, while Inertial Measurement Units (IMUs) measure the orientation of the user to generate an audio signal that makes the distance as well as the orientation of obstacles audible. This application consists of multiple parallel tasks that are mapped to an ASIR-based MPSoC. Both applications show the performance advantages of ASIRs compared to a conventional NoC-based MPSoC. Furthermore, dynamic partial reconfiguration in terms of relocation and security aspects are investigated. The third major contribution refers to development and programming methodologies of NoC-based MPSoCs. A software-defined approach is presented that combines the design and programming of heterogeneous MPSoCs. In addition, a Kahn-Process-Network (KPN) –based model is designed to describe parallel applications for MPSoCs using ASIRs. The KPN-based model is extended to support not only the mapping of tasks to NoC-based MPSoCs but also the mapping to ASIR-based MPSoCs. A static mapping methodology is presented that assigns tasks to ASIRs and processors for a given KPN-model. The impact of external hardware components such as sensors, actuators and accelerators connected to the processors is also discussed which makes the approach of high interest for embedded systems.