Montgomery reduction inline asm revisited

[kilic]

This is biproduct of #49 review. So similarly montgommery reduction is written with double carry chain. montgomery_reduce_short function is also implemented as @jonathanpwang suggested. For 1M sample to_repr function performs like below:

new to_repr ..................................................................29.586ms
old to_repr ..................................................................35.485ms

[CPerezz]

The changes look good to me. It would be nice anyways to add some kind of source link or documentation in regards the algorithm that is being implemented.

As otherwise, with time, it will be challenging to refactor this if we don't know where it came from.

[han0110]

LGTM! The assembly is a little bit different than https://github.com/mratsim/constantine/blob/151f284/constantine/math/arithmetic/assembly/limbs_asm_redc_mont_x86_adx_bmi2.nim#L187-L230 but the logic seems same to me, not sure how the #cycle compares tho.

Also it'd be nice to rebase to check with the latest CI, it's weird that I couldn't compile locally (need to move the in(reg) before out(...) to compile), not sure what's going on here.

[han0110]

Leftover?

[han0110]

Leftover?

[han0110]

Leftover?

[kilic]

LGTM! The assembly is a little bit different than https://github.com/mratsim/constantine/blob/151f284/constantine/math/arithmetic/assembly/limbs_asm_redc_mont_x86_adx_bmi2.nim#L187-L230 but the logic seems same to me, not sure how the #cycle compares tho.

Also it'd be nice to rebase to check with the latest CI, it's weird that I couldn't compile locally (need to move the in(reg) before out(...) to compile), not sure what's going on here.

Thanks. Let me check against constantine impl before the merge

[CPerezz]

Any news @kilic ??

[kilic]

Also it'd be nice to rebase to check with the latest CI, it's weird that I couldn't compile locally (need to move the in(reg) before out(...) to compile), not sure what's going on here.

Remote CI also complained so it is fixed too.

Any news @kilic ??

Now it should be ready to merge