Why is there a noticeable change in P-1 speed between 107M and 109M exponents

[ewmayer]

Originally Posted by Uncwilly

Since the software is using the Floating Point Unit of the CPU, there is that noise at the end. Which is ok for most normal calculations as it way below what the user needs. But since we are using nearly the entire width of the instruction set, we care about the accumulated errors.

ITYM "width of the floating-point mantissa".

For IEEE64 math and world-class-accurate bignum code, the crossover from 5.5M-6Mdouble FFT is right around 107Mbit.