What’s the impact of polyhedral compiler techniques on downstream code-writers?

I’ve seen a lot of buzz and literature on polyhedral compilers recently. For a SWE or scientific computing person who uses compilers but doesn’t care too much for their inner workings, what kinds of impacts should one inspect?

Do polyhedral techniques reduce compile time? Do they enable new optimizations that weren’t possible before? Do they make compilers easier to write/ maintain?

Specific examples appreciated.