Questions about mpn_powm

[? ??]

Hello, I am a student developing a library for modular arithmetic, and find out that your function “mpn_powm” uses REDC reduction and sliding window algorithm, so does mine, but your function is much faster than mine(10 times faster for a 256-bit prime). I see your source code of mpn_powm, but cannot find something special. The attachment is my code, which is much shorter, with a correct output. I wonder how you make it so fast, thanks!

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: pow.c
URL: <https://gmplib.org/list-archives/gmp-discuss/attachments/20190413/b409e451/attachment.c>