Perhaps the performance difference is due to resource limitations in the Shakti trial?