List Commands Library for 50g

03272018, 10:07 PM
Post: #306




RE: List Commands Library for 50g
(03252018 05:52 PM)DavidM Wrote: A quick check shows that it is definitely cycling the internal RAND seed, perhaps twice per element. I'll check into it further. Finally got around to this... Here's the pertinent subroutine that is called to generate each random entry: Code: (FPTR 3 54=3:6AA24) ...so it's essentially the same method used by 99% of random generators out there. As John mentioned, the second %RAN call is for the sign of the result. No surprises there. FWIW, I'm now experimenting with an alternative approach to selecting the swap target for the FisherYates shuffle in LSHUF:
Performance has been good with this new method, and may even be faster than the previous one since there are fewer conversions taking place at each iteration (all the local data stays in binary form). Would still love to hear thoughts on reasonable ways to test the results... 

« Next Oldest  Next Newest »

User(s) browsing this thread: 1 Guest(s)