incorrectly multiplies the high 32 bits of the DECIMAL by 1e64 instead
of the correct 2^64.