The evaluation of slope stability by variational calculus can give non-univocal results, what is provoked both by specific form of the variational solution and accumulation of rounding and truncation errors during integration processes. I decided to apply a genetic algorithm with advanced techniques of information processing to manage the whole finding procedure. To increase the efficiency of computations a concept of parallelism was exploited, using metacomputer composed of 10 PC handled by Parallel Virtual Machine. Unfortunately, I could not utilize our Computer Laboratory only for my own purposes (students have priority :) ), so I contrived an idea of multiprocessor machine, cheap and easy to build. So, twenty computational cells were built, each cell composed of three one-chip microcomputers: two 89C2051 for communication and buffering and one LPC2106 working as a main calculation core.
A few pictures of single cell and cellputer are presented below.
Scheme of working cell.
Template of working cell.
View of working cell.
Scheme of cellputer.
View of connected cells.
View of cellputer (20 cells).
