Pops an i64 value off the stack, which must be the binary form of a f64 floating point number, copy the bits into a f64 value, and pushes it onto the stack. The binary structure of a f64 value is copied as it is, not its value, from an i64 value. This is like copying the floating point value into memory and then reading it with an integer. The binary bits are the same, however the numeric value is totally different.

f64.reinterpret_i64

;; Translate f64 1234.567 into an i64 value
f64.const 1234.567
i64.reinterpret_f64

;; The stack contains an i64 value of the binary f64 value for 1234.567

;; Translate it back from the i64 binary into a f64 value
f64.reinterpret_i64

;; The stack contains a f64 value of 1234.567