Most work on parallel GAs has been concerned with extracting the largest possible speedup assuming a well defined, static, dedicated parallel machine architecture. This does not meet the needs of the practitioner who wishes to parallelise a GA across a network of existing machines in an organisation. In this case the practitioner is more likely to be concerned about generality, and temporal and spatial robustness, rather than purely maximizing efficiency.

Parametric parallelisation of a GA using Clustor is relatively simple to setup and is fault tolerant. This project dealt with a large problem whose objective function was sufficiently complex to yield a high computation to communication ratio. For this type of problem, parallelisation through parameterization is a simpler and more robust option than message passing.