Title :
Profiling the fault tolerance for the adaptive Protein Processing Associative Memory
Author :
Qadir, Omer ; Timmis, Jon ; Tempesti, Gianluca ; Tyrrell, Andy
Author_Institution :
Dept. of Electron., Univ. of York, York, UK
Abstract :
The Protein Processor Associative Memory (PPAM) is a novel hardware architecture for a distributed, decentralised bidirectional, hetero-associative memory, that can adapt online to changes in the training data. The PPAM is fundamentally different from traditional processing methods that tend to use arithmetic operations to perform computation. In this paper, we evaluate the fault tolerant properties of the PPAM and show that, given a sufficiently large number of Conflict-Resolving nodes in the network, faults result in graceful degradation and individual nodes are not critical to the correct operation of the whole.
Keywords :
content-addressable storage; fault tolerance; PPAM; adaptive protein processing associative memory; arithmetic operations; conflict-resolving nodes; decentralised bidirectional memory; distributed memory; fault tolerance profiling; graceful degradation; hardware architecture; hetero-associative memory; Circuit faults; Computer architecture; Delay; Fault tolerant systems; Hardware; Redundancy;
Conference_Titel :
Adaptive Hardware and Systems (AHS), 2012 NASA/ESA Conference on
Conference_Location :
Erlangen
Print_ISBN :
978-1-4673-1915-7
Electronic_ISBN :
978-1-4673-1914-0
DOI :
10.1109/AHS.2012.6268658
