Branch prediction and speculative execution consist of making probabilistic predictions about the likely near-term evolution of the near-Earth space, and distributing among the cluster machines simulations that assume each of the probabilistically predicted outcomes as initial conditions. As the near-Earth space evolves and real-time satellite data get assimilated into the algorithm, some of the speculatively executed simulations will be proved wrong. At that point the machines that were executing them will be reassigned either to new lines of speculative simulation, or to increase the processing power devoted to more promising simulations already executing.

Branch prediction and speculative execution have been exceedingly successful in the design of microprocessors, allowing CPUs to attain average processing speeds much higher than linear code execution would permit. The scheme is particularly suited to Space Weather for the same reasons that made it so successful in microprocessor design: our upwind early warning sentries can provide only sparse sampling of the incoming solar wind, while the bulk of our monitors, which can provide significantly better coverage, are located close to Earth and provide much shorter lead times. Estimating the different evolution branches of the tail and speculatively executing simulations along those branches, assures that the result of the simulations is already at hand by the time the denser near-Earth data becomes available to resolve the branch point.