single-float

Only supports single precision float.

CPUs: