Some ARM and x86 CPUs have instructions to convert between 16-bit and 32-bit float formats. Actually supporting arithmetic ops on half floats seems to be ...
確定! 回上一頁