An integral transport equation-based industrial radiography simulation code
is parallelized using the Message Passing Interface standard on computers
with both distributed- and shared-memory architectures. The algorithm invol
ves partitioning of the problem domain into regions that are connected to e
ach other through interface conditions. This results in a simultaneous set
of integral transport equations. Each equation in the set is assigned to a
different processor in the platform. The new algorithm is subjected to scal
ability tests in both cluster and shared-memory architectures for a varying
number of processors with different problem domain partition strategies. T
he results show a high level of scalability with favorable results in both
architectures.