i32 @__nv_mul24(i32 %x, i32 %y)
Calculate the least significant 32 bits of the product of the least significant 24 bits of x and y. The high order 8 bits of x and y are ignored.
Returns the least significant 32 bits of the product x * y.
Compute 2.0: Yes
Compute 3.0: Yes
Compute 3.5: Yes