National and Kapodistrian University of Athens
Fault detection methodology for caches in reliable modern VLSI microprocessors based on instruction set architectures
Abstract
dc:descriptionThe present PhD thesis introduces a low cost fault detection methodology for small embedded cache memories that is based on modern Instruction Set Architectures and is applied with Software-Based Self-Test (SBST) routines. The proposed methodology applies March tests through software to detect both storage faults when applied to caches that comprise Static Random Access Memories (SRAM) only, e.g. L1 caches, and comparison faults when applied to caches that apart from SRAM memories comprise Content Addressable Memories (CAM) too, e.g. Translation Lookaside Buffers (TLBs). The proposed methodology can be applied to all three cache associativity organizations: direct mapped, set-associative and full-associative and it does not depend on the cache write policy. The methodology leverages existing powerful mechanisms of modern ISAs by utilizing instructions that we call in this PhD thesis Direct Cache Access (DCA) instructions. Moreover, our methodology exploits the native performance monitoring hardware and the trap handling mechanisms which are available in modern microprocessors. Moreover, the proposed Methodology applies March compare operations when needed (for CAM arrays) and verifies the test result with a compact response to comply with periodic on-line testing needs. Finally, a multithreaded optimization of the proposed methodology that targets multithreaded, multicore architectures is also presented in this thesis.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2012
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Theodorou, Georgios
- Θεοδώρου, Γεώργιος
Subjects
dc:subject × 14- Έλεγχος μικροεπεξεργαστών
- Αξιοπιστία
- Ανίχνευση ελαττωμάτων που εφαρμόζεται με λογισμικό αυτοδοκιμής
- Έλεγχος εντός λειτουργίας
- Έλεγχος μνημών
- Microprocessor testing
- Reliability
- Software-based self-test
- On-line testing
- Memory testing
- Φυσικές Επιστήμες
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Natural Sciences
- Computer and Information Sciences
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/29098