mirror of
https://github.com/ROCm/ROCm.git
synced 2026-02-21 03:00:39 -05:00
clipping float8e4b15 to +-1.875 is a bad idea because these are represented as 0x7f and 0xff, which are +- nan on H100 for float8e4nv. We lose two values but this will make compatibility with float8e4nv way less painful. (it will just be a matter of adjusting the bias)