RPL mini-challenge: Create an anti-Identity Matrix

[DavidM]

(05-13-2018 02:29 AM)Claudio L. Wrote:  

(05-12-2018 08:04 PM)Zaphod Wrote:  1st version 5.5secs ish
2nd version 3 secs ish

Thanks for the timings. Ouch, very slow...

Claudio, I've tested your 2nd program on my 50g with much better results:

With exact integers (input, constants, and output): 1.62s
With approximate numbers (input, constants, and output): 0.33s

I'm not sure about the difference with Zaphod's timings. Perhaps he was using a different calculator? The above figures were the average of 10 iterations using 30 (or 30.) as the input, each iteration being prefaced by a GC that was excluded from the timing window.

I did have to make one small change, though: dropping the tick marks around K in the FOR loop declaration. That's probably a subtle difference between New- and Old-RPL.

Edit: We were posting at (nearly) the same time. The timings above are for the second program in your original post, not the most recent one you posted. I'll post an update after I run the most recent one under the same conditions. I think it will be slower, though.