DmytroTym

Hello @btilmon Pretty sure that his approach is faster than our default NTT. However we still lack support for Goldilocks in general (see https://github.com/ingonyama-zk/icicle/issues/90)

I closed a similar issue [here](https://github.com/ingonyama-zk/icicle/issues/16). I guess we can leave this one in case we want to implement Montgomery after all

@ImmanuelSegol not sure about this issue, is it the same as point 2 in https://github.com/ingonyama-zk/icicle/issues/137?

I feel like two different things are at play here: modular multiplication, which is what Montgomery multiplication does, and single-scalar multiplication, which is what double-and-add does. Admittedly, we have room...

To comment on this: we haven't implemented a more-efficient single-scalar multiplication than double-and-add yet, but we don't really have any use-cases for it so it's still low priority.

About arkworks: we put it under a feature making all arkworks-related crates optional dependencies. Not sure if that's good enough or do we still want to refactor it into a...