My daily work is math intensive and performance critical. I constantly need to think about cache sizes, big-O scaling and that ever-important constant factor.

My daily work is math intensive and performance critical. I constantly need to think about cache sizes, big-O scaling and that ever-important constant factor. Me too, at least for some of the projects I'm working on, but it's probably fair to say that I've never used most of the "textbook algorithms" I've come across over the years in a production system. The interesting and challenging problems are mostly the ones that aren't just a direct application of such an algorithm. Consequently, I think the more valuable ability is recognising when a part of such a problem probably does have a standard solution, so you can spend a few minutes looking up the current state of the art. Obviously you still need enough familiarity with the theory in the areas you're working in for that recognition to work, and then to understand the current state of the art when you need it.