You’re looking at noise. First, use perf_counter
instead of process_time
; second, jack up the size of a
by 100x. Then the results are repeatable and make sense (to me anyway).
You’re looking at noise. First, use perf_counter
instead of process_time
; second, jack up the size of a
by 100x. Then the results are repeatable and make sense (to me anyway).