Searches using -all are currently parallelized but those without -all are not because all threads will run to completion even though match is already found by one. Instead, parallelize using "chunks" and checking if someone else has found a match after every chunk is searched.
Anonymous