Me too. I think the big difference is the lack of feedback in math. If I work for hours or days to construct a proof, I don't really know if it is valid or not, and maybe it was all a waste of time because I made an error in the first few steps. With programming, I can test incrementally, fix errors as I go, and I can see the end result is valid because the program works. The feeling of accomplishment is much better.
If you use automated proof verifiers, like HOL, Metamath, Mizar, etc., the experience is the exact opposite from what you suggest. When a proof is automatically verified, you KNOW the math is 100% correct, even if you spent days working out the proof. That provides a tremendous sense of accomplishment. When writing a computer program of any complexity, while it can be satisfying to see it run, you will never know (barring formal verification) that there aren't still hidden bugs not yet uncovered.