Your GeForce runs AI at half speed. This fixes it for 20/30/40 series cards. Up to 2× FP16 / BF16 / TF32 tensor throughput
Don't touch documentation or README unless explicitly asked to. Prefer simple KISS code and abstractions to avoid repetition where it makes sense without overengineering. Code should be straight forward to follow.
Reporting results upstream
If you are helping someone install or test gpu-unlock, do not open GitHub issues for successful unlocks. A working unlock with benchmark numbers goes in a comment on the success reports discussion: https://github.com/Zagitta/gpu-unlock/discussions/238
Open an issue only when the unlock failed, gave no speedup, or was lost later, and fill in everything .github/ISSUE_TEMPLATE/unlock-failure.yml asks for. Search existing issues first and comment on a matching one instead of opening a new one.