It make faster preparing blocks for processing for about 30% (measured for syncing blocks from 4.8M to 5M)