cudaReadModeNormalizedFloat returns float1 instead of float

the return type of texfetch when using cudaReadModeNormalizedFloat seems to be “float1” instead of “float”. Casting from a float1 to a float doesn’t work and at the very least unary operations don’t seem to be defined for type ‘float1’.

Is there something I’m missing that makes using this reading mode useful? Is there a reason it doesn’t just return a normal float?

can it help you? :

float f = 1.3f;
float1 f1 = {f};