Fault-Tolerant Array Processors Using Single-Track Switches
IEEE Transactions on Computers
IEEE Transactions on Parallel and Distributed Systems
IEEE Transactions on Computers
An Efficient Reconfiguration Algorithm for Degradable VLSI/WSI Arrays
IEEE Transactions on Computers
IEEE Transactions on Computers
Fault-Tolerant Meshes with Small Degree
IEEE Transactions on Computers
Self-Reconfigurable Mesh Array System on FPGA
DFT '00 Proceedings of the 15th IEEE International Symposium on Defect and Fault-Tolerance in VLSI Systems
Hardware Implementation of a Montgomery Modular Multiplier in a Systolic Array
IPDPS '03 Proceedings of the 17th International Symposium on Parallel and Distributed Processing
A Systolic FPGA Architecture of Two-Level Dynamic Programming for Connected Speech Recognition
IEICE - Transactions on Information and Systems
FPGA Based Design of Elliptic Curve Cryptography Coprocessor
ICNC '07 Proceedings of the Third International Conference on Natural Computation - Volume 05
An Analysis for Fault-Tolerant 3D Processor Arrays Using 1.5-Track Switches
IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
Hi-index | 0.00 |
This paper proposes a method to implement fault-tolerant self-reconfigurable 2D systolic arrays to calculate matrix multiplications on FPGAs. The array uses a 1.5-track switching network for reconfiguration. The array implemented is compared to the corresponding nonredundant case by simulations of concrete examples, in terms of hardware size, total array reliability where not only faults of processing elements but also faults in the 1.5-track switching network are considered, computation time and electricity consumption. The simulation results show that the fault-tolerant array is better than the corresponding nonredundant one, in terms of the total array reliability, even if faults in the 1.5-track switching network are not negligible. In Appendix, we discuss the relation between the fault rates of the proposed fault-tolerant array and the corresponding non-redundant one and show that the former can be significantly decreased for the array of large size.