Subscribe
Issue No.09 - September (1997 vol.46)
pp: 997-1015
ABSTRACT
<p><b>Abstract</b>—<it>Structural fault tolerance (SFT)</it> is the ability of a multiprocessor to reconfigure around faulty processors or links in order to preserve its original processor interconnection structure. In this paper, we focus on the design of SFT multiprocessors that have low switch and link overheads, but can tolerate a very large number of processor faults on the average. Most previous work has concentrated on deterministic <it>k</it>-fault-tolerant (<it>k</it>-FT) designs in which exactly <it>k</it> spare processors and some spare switches and links are added to construct multiprocessors that can tolerate any <it>k</it> processor faults. However, after <it>k</it> faults are reconfigured around, much of the extra links and switches can remain unutilized. It is possible within the basic node-covering framework, which was introduced by Dutt and Hayes as an efficient <it>k</it>-FT design method, to design FT multiprocessors that have the same amount of switches and links as, say, a two-FT deterministic design, but have <it>s</it> spare processors, where <tmath>$s \gg 2,$</tmath> so that, on the average, <it>k</it> = Θ(<it>s</it>) (<it>k</it>≤<it>s</it>) processor failures can be reconfigured around. Such designs utilize the spare link and switch capacity very efficiently, and are called <it>probabilistic</it> FT designs. An elegant and powerful method to construct covering graphs or CG's, which are key to obtaining the probabilistic FT designs, is to use linear error-correcting codes (ECCs). We show how to construct probabilistic designs with very high average fault tolerance but low wiring and switch overhead using ECCs like the 2D-parity, full-two, 3D-parity, and full-three codes. This design methodology is applicable to any multiprocessor interconnection topology and the resulting FT designs have the same node degree as the non-FT target topology. We also analyze the deterministic fault tolerance for these designs and develop efficient layout strategies for them. Finally, we compare the proposed probabilistic designs to some of the best deterministic and probabilistic designs proposed in the past, and show that our designs can meet a given mean-time-to-failure (MTTF) specification at much lower hardware costs (switch complexity, redundant wiring area, and spare-processor overhead) than previous designs. Further, for a given number of spare processors, our designs have close-to-optimal reconfigurabilities that are much better than those of previous probabilistic designs.</p>
INDEX TERMS
Average fault tolerance, deterministic fault tolerance, fault-tolerant multiprocessors, linear error-correcting codes, matching, network flow, node-covering, VLSI layout, reconfiguration.
CITATION
Shantanu Dutt, Nihar R. Mahapatra, "Node-covering, Error-correcting Codes and Multiprocessors with Very High Average Fault Tolerance", IEEE Transactions on Computers, vol.46, no. 9, pp. 997-1015, September 1997, doi:10.1109/12.620481