Fun with Numbers: The PanPrimeDigit Cube Hypothesis

08112017, 10:26 PM
Post: #35




RE: Fun with Numbers: The PanPrimeDigit Cube Hypothesis
(08112017 04:43 PM)DavidM Wrote: I think I'll start with my original plan, which was to use a Saturn routine to cube/validate the last 5 digits of the current root before proceeding to check the entire root. That will exclude a fairly large percentage of the possible targets (about 98+% I believe). Even with that, I'm thinking a runtime >24hrs is likely. Probably should have done this simple test before posting the above, but thought I'd share this now that I have: Timing a simple SysRPL loop that simply checks a loop counter shows a speed of about 2100 cycles/sec on a real 50g. While that seems fairly fast, it isn't fast enough to pursue the abovementioned strategy for this application. Knowing in advance that the knowntarget is at the 1405349897th loop iteration, that implies a minimum of over 7 days worth of cycles just for the looping structure itself. The actual root/cube checking would add considerable more time, so I won't pursue that method any further. I'll probably try a modified (lower "order") form of Werner's approach, and I may also go ahead and put together a Saturn routine for testing an exact integer to see how long it takes. That will help define whether this is worth pursuing further on this platform. Either way, it's been an interesting exercise. Thanks Joe! 

« Next Oldest  Next Newest »

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