[K0SM0S]

~5 years. Medium-sized team in-house + hordes (hundreds, thousands) of engineers in the field helping clients on-site, writing code for them directly upstreamed to drivers, core libs, etc. (iteratively optimized in-house, ship feature, rinse and repeat). Story of the PlayStation SDKs, of DX too, but above all CUDA (they really outdid this strategy), now for cuDNN and so much more.

It takes incompressible time because you have to explore the whole space, cover most bases; and it takes an industry several years (about one "gen" / hardware cycle) to do that meaningfully.It helps when your platform is disruptive and customers move fast.

Maybe 3 years at best if you start on a new ideal platform designed for it from scratch. And can throw ungodly amount of money fast at it (think 5K low-level engineers roaming your installed base).

Maybe 10+ yrs (or never) if you're alone, poor, and Radeon (j/k but to mean it's non-trivial).