3.190. __nv_half2float

float @__nv_half2float(i16 %h)

Convert a half-precision float to a single-precision float in round-to-nearest-even mode.

Convert the half-precision floating point value x represented in unsigned short format to a single-precision floating point value.