Threaded Mode | Linear Mode

FLISZT · 12-17-2022, 02:31 AM

(12-13-2022 02:37 AM)Joe Horn Wrote: Here's a brute-force implementation of that method:

LBL A, 0, STO I, Rdn, LBL 0, g ISZ, f SR, g x≠0, GTO 0, RCL I, RTN

GSB A in any binary mode returns the number of bits needed to express X.

It works, but it's horrifically slow:
Input 2^40-1 outputs correct result of 40 in 15 seconds.
Input 2^40 outputs correct result of 41 in 15 seconds.

EDIT: 0 NOT with WSIZE of 64 returns 64 in 22 seconds.

That's not that much for a calculator that came out 40 years ago. My Sanyo (1977) needs more than 30s to compute the factorial of 69!!
I should have bought one 16C when you could find it in stores… Of course, with a DM16L, it would be faster (and it's about half the price).

(12-13-2022 09:17 AM)Joe Horn Wrote: Here's a 16C-only routine that doesn't require any looping. The LJ command is on the [g] [A] key. #B is on the [g] [7] key.

LJ, 0, NOT, #B, x<>y, –

As before, put any value in X in any binary mode. The above routine returns the number of bits required to express X. If WSIZE is sufficiently large to put X on the stack, then this routine will always return the correct answer.

Inputs of 2^40-1 and 2^40 return the correct answers (40 and 41 respectively) in roughly 1.5 seconds, with a speed hit for a large WSIZE.

I have tried both of your programs at: https://stendec.io/ctb/rpn_prog.html
Thank you! Smile

Too bad the RPL calculators don't have all the functions of the 16C. However, trying to program the missing functions might be a good exercise!