Semiconductor device modeling on a workstation
We choose to move from large mainframe computers to workstations to gain the interactive graphics we need to prepare and to analyze semiconductor device modeling problems. Given this much on a workstation, it is convenient to attempt to solve the entire problem there. We find that a top-of-the-line Apollo 660 workstation, with bit-slice processor, pipelined arithmetic processor, and 4 megabytes of real memory, is surprisingly effective in finding solutions when running the Pisces II device modeling code. In our experiment we find where the workstation bogs down when running these problems. We both analyze the Pisces CPU time log and we sample the executing program to accumulate a histogram of execution time as distributed over the source code. Results suggest how Pisces could be adapted to solve somewhat larger problems entirely on the workstation. Evolution of a trusted derivative of Pisces, to be used on supercomputers without interactivity, is suggested to complement our success with Pisces on workstations. 4 refs.
- Research Organization:
- Sandia National Labs., Albuquerque, NM (USA)
- DOE Contract Number:
- AC04-76DP00789
- OSTI ID:
- 5975655
- Report Number(s):
- SAND-85-0844; CONF-8504207-1; ON: DE86001125
- Resource Relation:
- Conference: Workshop on large scale computational device modeling, Chicago, IL, USA, 18 Apr 1985
- Country of Publication:
- United States
- Language:
- English
Similar Records
Data Locality Enhancement of Dynamic Simulations for Exascale Computing (Final Report)
Institute for Sustained Performance, Energy, and Resilience (SuPER)