Title :
Parallel 3-D MOSFET simulation
Author :
Bagheri, Babak ; Ilin, Andrew ; Scott, L. Ridgway
Author_Institution :
TCAMC, Houston Univ., TX, USA
Abstract :
Describes a parallel solver for the nonlinear Poisson-Boltzmann equation modeling the distribution of an electrostatic potential in a MOSFET in three dimensions. The nonlinear system is solved with Newton´s method. The linearized mathematical model is solved by the conjugate gradients method preconditioned by a parallel version of incomplete Cholesky factorization. This technique is easy to implement and reasonably scalable. With this method, one can solve the nonlinear Poisson-Boltzmann equation on a grid of size 160/sup 3/ in approximately 15 minutes on 16 processors of a KSR1 multiprocessor. This time reflects a superlinear speedup of 19 for 16 processors. The greatest barrier to scalability appears to be related to having processor subdivision boundaries cutting across areas where the grid is refined most and the solution is highly variable.<>
Keywords :
conjugate gradient methods; digital simulation; electronic engineering computing; insulated gate field effect transistors; parallel algorithms; semiconductor device models; 3-D MOSFET simulation; Newton´s method; conjugate gradients method; electrostatic potential; multiprocessor; nonlinear Poisson-Boltzmann equation; nonlinear system; parallel solver; scalability; semiconductor device simulation;
Conference_Titel :
System Sciences, 1994. Proceedings of the Twenty-Seventh Hawaii International Conference on
Conference_Location :
Wailea, HI, USA
Print_ISBN :
0-8186-5090-7
DOI :
10.1109/HICSS.1994.323186