Looks like DeepMind's new sorting algorithm is quite similar to those tried in that question
If we adapt it to the settings of the SO question, how does it actually compare to what humans proposed?
Asked
Active
Viewed 38 times
0
-
2I think calling its modification *a new sorting algorithm* hyperbole. It looks similar to what the GNU "superoptimizer" did (does?). It is not *discovering a new algorithm*, but *implementing a worth-while micro-optimisation*. – greybeard Jun 24 '23 at 12:36
-
Yes, but that was also what we users of SO basically did in the linked question. Which could lead to some interesting comparisons (I'm also wondering why posters did not found that specific micro optimization as the answers were quite thorough). – kriss Jun 24 '23 at 18:03