The computation time estimates of the seismic data processing parallel programs depending on the number of computing processes and the data distribution strategy are proposed. The estimates help to optimize the parallel program execution in the specific software/hardware architecture on the base of few experiments as well as to fit optimal architecture for solving typical seismic data processing problems.