RPL Mini-Challenge: All Odd Digits?

[DavidM]

(07-03-2017 07:52 AM)Gilles59 Wrote:  Very interesting David!

I dont have my 50G here, but I think there is a way to improve speed of the part :

Code:

 R\->I ALOG 1 - 9 / \->STR

I think at something like :

Code:


 1. - R\->I ALOG  \->STR "0" "1" SREPL DROP

I tried some variants of the above, but found that the version I posted ended up being faster, at least on a real 50g vs. Emu48. This is one of those cases where the performance on Emu48 is misleading when compared to a real 50g. When I use your code instead of the first version posted, the overall speed on a real 50g goes from 0.1222 seconds to around 0.167. The SREPL version is faster on Emu48, though, probably because Emu48 doesn't emulate the ARM layer. So I guess it depends on what platform you're targeting.

Unfortunately the SREPL version is also about 15 bytes longer, so it not only slows down but also raises the score on that front as well. I've been assuming that Joe's scoring method is based on the results of size/performance as measured on a real calculator as opposed to an emulated one.