Performance of modular exponentiation

[Ivo maffei]

Hi everyone,

I’m a DPhil student at the university of Oxford and I’m looking at parallelising modular exponentiation (mpz_powm).
I’m using the GMP 6.2.1 release for C and it is really fast. Could someone explain me how GMP achieves this performance?
If anyone has a somewhat detailed documentation of the code, that would also be very useful as the source code is quite convoluted.

Thanks a lot for any help.
Ivo