fp16fml

Enable full half-precision floating point fml instructions.

CPUs: