testing should fix one variable

instead of making each run by using different seeds, I should use the same seed for different runs, since the data and block size keep changing, which are the parameters I care about, so the lesson is: fix those uninterested things.