News for hackers. Stuff that matters.
Re:Compiler Optimizations are Awesome by dagw on Thursday June 01, 2017 @06:21AM (#14459135)
You may be significantly underestimating your ability to get phenomenal gains from algorithmic improvements.

My HPC lecturer used to say that the speed difference between bad Matlab code and good Matlab code will often be greater than the speed difference between good Matlab code and good Fortran code.

Tweaking the hell out of your code at the assembly level is a waste of time if you're using a naive O(2^N) algorithm.

Re:Compiler Optimizations are Awesome by sllabres on Thursday June 01, 2017 @10:21AM (#14460433)
You mean O(n^2). 2^n looks a more than 'naive' for most cases.
Re:Compiler Optimizations are Awesome by gpderetta on Thursday June 01, 2017 @09:31AM (#14460053)
... unless N is small.