What constraints/coordination exists with this, in terms of host driver support? What enforces that Nix does not attempt to use a newer cuda toolkit on a host with an older cuda driver?
You pin the CUDA toolkit version compatible with your driver; manifest.lock ensures zero drift.
Driver version is host-managed (stable), toolkit is hash-pinned (stable). No drift on either side.
Initial selection matters (pick compatible versions, nothing automatic will stop you from initially installing the wrong thing), but once pinned, and unless/until you change it, you get the same toolkit forever regardless of catalog updates or time passing.
Driver version is host-managed (stable), toolkit is hash-pinned (stable). No drift on either side.
Initial selection matters (pick compatible versions, nothing automatic will stop you from initially installing the wrong thing), but once pinned, and unless/until you change it, you get the same toolkit forever regardless of catalog updates or time passing.