f33-book-depend-pres-pt5

f33-book-depend-pres-pt5 - Robust Parallel Processing...

Info iconThis preview shows pages 1–9. Sign up to view the full content.

View Full Document Right Arrow Icon
Nov. 2009 Part V – Malfunctions: Slide 1 Robust Parallel Processing Resilient Algorithms
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Nov. 2009 Part V – Malfunctions: Slide 2 About This Presentation This presentation is intended to support the use of the textbook Dependable Computing: A Multilevel Approach (traditional print or on-line open publication, TBD). It is updated regularly by the author as part of his teaching of the graduate course ECE 257A, Fault-Tolerant Computing, at Univ. of California, Santa Barbara. Instructors can use these slides freely in classroom teaching or for other educational purposes. Unauthorized uses, including distribution for profit, are strictly prohibited. © Behrooz Parhami Edition Released Revised Revised Revised Revised First Sep. 2006 Oct. 2007 Nov. 2009
Background image of page 2
Nov. 2009 Part V – Malfunctions: Slide 3 17 Malfunction Diagnosis
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Nov. 2009 Part V – Malfunctions: Slide 4
Background image of page 4
Nov. 2009 Part V – Malfunctions: Slide 5 Robust Parallel Processing Resilient Algorithms
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Nov. 2009 Part V – Malfunctions: Slide 6 17.1 Self-Diagnosis in Subsystems Layered approach: A small part of a unit is tested, which then forms a trusted core The trusted core is used to test the next layer of subsystems Region of trust is gradually extended, until it covers the entire unit One approach to go/no-go testing based on self-diagnosis Tester supplies a random seed to the built-in test routine The test routine steps through a long computation that exercises nearly all parts of the system, producing a final result The tester compares the final result to the expected result Ideally, if a properly designed self-test routine returns a 32-bit value, the value will match the expected result despite the presence of faults with probability 2 –32 10 –9.6 test coverage = 1 – 10 –9.6 Core Self-diagnosis initiation Unit A Unit B Unit C
Background image of page 6
Nov. 2009 Part V – Malfunctions: Slide 7 17.2 Malfunction Diagnosis Model Diagnosis of one unit by another The tester sends a self-diagnosis request, expecting a response The unit under test eventually sends some results to the tester The tester interprets the results received and issues a verdict Testing capabilities among units is represented by a directed graph i Tester j Testee Test capability I think j is okay (passed test) 0 i Tester j Testee Test capability I think j is bad (failed test} 1 The verdict of unit i about unit j is denoted by D ij {0, 1} All the diagnosis verdicts constitute the n × n diagnosis matrix D The diagnosis matrix D is usually quite sparse M 0 M 1 M 3 M 2
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Nov. 2009 Part V – Malfunctions: Slide 8 More on Terminology and Assumptions Malfunction diagnosis in our terminology corresponds to “system-level fault diagnosis” in the literature The qualification “system-level” implies that the diagnosable units are subsystems with significant computational capabilities (as opposed to gates or other low-level components) We do not use the entries on the main diagonal of the diagnosis matrix D (a unit does not judge itself) and we usually do not let two units test one another
Background image of page 8
Image of page 9
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 72

f33-book-depend-pres-pt5 - Robust Parallel Processing...

This preview shows document pages 1 - 9. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online