DocumentCode :
2391813
Title :
A new matrix solution technique for large-scale circuit simulation on multi-processor system
Author :
Jia, X.D. ; Chen, R.M.M.
Author_Institution :
Dept. of Electron. Eng., City Polytech. of Hong Kong, Kowloon, Hong Kong
fYear :
1994
fDate :
22-26 Aug 1994
Firstpage :
832
Abstract :
In this paper a new matrix solution technique is proposed for large-scale circuit simulation. This technique combines direct and iterative relaxation-based methods and is very suitable for parallel processing. In order to make the technique robust or generally applicable, a heuristic method to identify the right couplings and strong feedbacks among the sub-circuits is proposed. The identification process is done by setting a tolerance for each equation. The tolerance computation only requires forward and backward substitution and can be done in parallel. A simple example to illustrate this technique is also given in this paper
Keywords :
circuit CAD; circuit analysis computing; digital simulation; matrix algebra; multiprocessing systems; backward substitution; direct iterative relaxation-based methods; feedback; forward substitution; heuristic method; identification process; iterative relaxation-based methods; large-scale circuit simulation; matrix solution technique; multiprocessor system; parallel processing; subcircuits; tolerance computation; Circuit simulation; Convergence; Coupling circuits; Equations; Feedback circuits; Iterative methods; Large-scale systems; Parallel processing; Robustness; Sparse matrices;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
TENCON '94. IEEE Region 10's Ninth Annual International Conference. Theme: Frontiers of Computer Technology. Proceedings of 1994
Print_ISBN :
0-7803-1862-5
Type :
conf
DOI :
10.1109/TENCON.1994.369194
Filename :
369194
Link To Document :
بازگشت