An efficient waveform iterative technique for device transient simulation on parallel distributed memory computers

In this paper, we mainly study the parallelization aspects of the accelerated waveform relaxation algorithms for the transient simulation of semiconductor devices on parallel distributed memory computers since these method are competitive with standard pointwise methods on serial machines, but are significantly faster on parallel computers. Here we propose an efficient variant of GMRES method, for solving the resulting sequence of time-varying sparse linear differential-algebraic initial-value problems (IVP) arising at each linearization step with waveform Newton. We will describe a more efficient alternative which avoids the global communication of inner products and only requires local communications on for massively distributed memory computers since the number of inner products represents the bottleneck of parallel performance. Experimental results carried out on Parsytec GC/PowerPlus with regards to the comparison with other accelerated approaches such as convolution SOR and waveform GMRES techniques on waveform relaxation algorithm and pointwise methods are described.