N-Queens Benchmark On 38E

[Dave Britten]

It turns out that the 38E and the 12C have effectively identical programming capabilities (minus the 12C's added financial functions like bonds and depreciation), and that includes the (mis)use of CFj and Nj to do indirect addressing, so you can use the 12C version of the benchmark program as-is on a 38.

See here for program listings:

https://www.hpmuseum.org/cgi-sys/cgiwrap...i?read=700

There is one minor stylistic issue to note with the 12C program listing: STO g Nj isn't actually a valid key sequence on either model. What happens is that the g cancels out the previous STO prefix, and you're just executing Nj. That's fine, because that's how you store a value. Pressing STO g Nj on the 38 puts the keycode 25 15 into program memory, which is just g Nj.

Anyway, the run time that I got on my 38E was 1h06m12s. That means the 12C is only about 15-16% faster than the 38E. I'd be curious to know if the 38C differs significantly.

And for anybody that was questioning the utility of the D.MSΔ% function I put in my percentage program for the 65, there you go.