Refine
Document Type
- Doctoral thesis (2)
- Habilitation thesis (1)
Has Fulltext
- yes (3)
Is part of the Bibliography
- no (3)
Keywords
- Selbstreparatur (3) (remove)
Institute
This thesis introduces a comprehensive approach for making a particular class of embedded processors self-testing and self-repairing, such that a limited amount of permanent hardware faults that occur during the lifetime of these processors in the field will not prohibit the functional behavior of the user application running on the processor. The presented concepts all use redundant hardware, but the techniques used for administrating the hardware-redundancy range from hardware-based methods over hybrid methods to pure software-based methods, whereby the focus is on the latter ones. The proposed methods will be demonstrated by using a processor that is well designed for diagnostic self-test and self-repair purposes. This will also highlight some architectural properties of such a processor, which are beneficial for performing a software-based self-test and self-repair process.
Chapter 1 is an introduction to the field of dependable systems and fault tolerance. Fundamental terms and notations, which are used throughout this thesis for classification and evaluation, are provided. The used processor model – the VARP processor – is introduced in chapter 2 together with a hardware-based self-repair scheme for that processor. The results are used as reference values for evaluating the software-based methods. Chapter 3 introduces the fundamental concept of the software-based self-repair. In chapter 4 hybrid methods are derived by combining software-based and hardware-based methods, highlighting the synergy effects of the combination. Finally, in chapter 5, a diagnostic and adaptive software-based self-test scheme is introduced. This self-test scheme provides the diagnostic capability that is needed in the field for identifying defect components in the VARP processor and completes the comprehensive software-based self-test and self-repair approach.
Die Skalierung von Bauelementen der Mikroelektronik bis hin zu atomaren Dimensionen hat einen zunehmenden negativen Einfluss auf deren Zuverlässigkeit und mittlere Lebensdauer. Durch uneinheitliches Skalieren der räumlichen Ausdehnung im Vergleich zur Versorgungsspannung steigen die internen Energiedichten und damit die Temperatur. Diese trägt wesentlich zur Verstärkung gewisser Fehlereffekte bei. Maßnahmen wie Selbstreparatur und Entstressen können dieser Entwicklung entgegenwirken, wobei Kombinationen solcher Ansätze noch bessere Ergebnisse versprechen. Ziel der vorliegenden Dissertationsschrift ist es, eine dieser Kombinationen auf ihre Kosten und Nutzen zu untersuchen. Damit kann am Ende die Aussage getroffen werden, ob die Activity Migration eine gewinnbringende Ergänzung in einem zur Selbstreparatur genutzten M aus N System darstellt oder nicht.
Um dieses Ziel zu erreichen, werden zuerst fünf verschiedene Implementierungen des Ansatzes in einem VLIW Prozessor erstellt. Anschließend findet die Simulation der Implementierungen in verschiedenen Strukturgrößen, in Bezug auf ihr Temperaturverhalten statt. Über entsprechende Modellierungen wird daraufhin die Zuverlässigkeit und mittlere Lebensdauer der Originalsysteme, Systeme mit reinem M aus N System und den Systemen mit integrierter Activity Migration in Abhängigkeit von Temperatur und Fläche ermittelt. Die erzeugten Ergebnisse belegen, dass das Hinzufügen der Activity Migration bei den untersuchten Prozessormodellen, mit starker thermischer Kopplung zwischen Funktions- und Ersatzbaugruppen, warmer Redundanz, Rekonfiguration im laufenden Betrieb und den gegebenen Formeln zur Berechnung der Fehlerraten verschiedener Fehlermodelle, keinen nennenswerten Gewinn an Lebensdauer erbringt, da das reine M aus N System bereits zu einer deutlichen Lebensdauersteigerung führt.
The reliability of interconnects on integrated circuits (IC) has become a major problem in recent years because of the rise of complexity, the low-k-insulating material with reduced stability, and wear-out-effects from high current densities. The total reliability of a system on a chip is increasingly influenced by the reliability of the interconnections, which is caused by increased communication from the elevated number of integrated functional units. In recent years, studies have predicted that static faults will occur more often decreasing the reliability and the mean time to failure. The most published solutions aim to prevent dynamic faults and to correct transient faults. However, built-in self-repair (BISR) as a solution for static faults has not previously been discussed along with the other possible solutions. Theoretically, BISR can lead to higher reliability and lifetime. This is my motivation to implement BISR for integrated interconnects. Because BISR cannot repair transient and dynamic faults, I combine BISR with other approved solutions in this thesis. The results show that the combination leads to higher reliability and lifetime with less area and static power overhead compared to the existing solutions.