How do I convert the first 32 bits of the fractional part of a Float into a Word32?

Say I have a Float. I want the first 32 bits of the fractional part of this Float? Specifically, I am looking at getting this part of the sha256 pseudo-code working (from the wikipedia)

# Note 1: All variables are unsigned 32 bits and wrap modulo 232 when calculating
# Note 2: All constants in this pseudo code are in big endian

# Initialize开发者_运维问答 variables
# (first 32 bits of the fractional parts of the square roots of the first 8 primes 2..19):
h[0..7] := 0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19

I naively tried doing floor (((sqrt 2) - 1) * 2^32), and then converting the returned integer to a Word32. This does not at all appear to be the right answer. I figured that by multipling by 2^32 power, I was effectively left shifting it 32 places (after the floor). Obviously, not the case. Anyway, long and the short of it is, how do I generate h[0..7] ?

The best way to get h[0..7] is to copy the hex constants from the Wikipedia page. That way you know you will have the correct ones.

But if you really want to compute them:

scaledFrac :: Integer -> Integer
scaledFrac x =
    let s = sqrt (fromIntegral x) :: Double
    in  floor ((s - fromInteger (floor s)) * 2^32)

[ printf "%x" (scaledFrac i) | i <- [2,3,5,7,11,13,17,19] ]