Skip to content
This repository has been archived by the owner on May 3, 2024. It is now read-only.

Prover optimization: memory reduction #77

Closed
Brechtpd opened this issue Apr 24, 2023 · 2 comments
Closed

Prover optimization: memory reduction #77

Brechtpd opened this issue Apr 24, 2023 · 2 comments

Comments

@Brechtpd
Copy link

Brechtpd commented Apr 24, 2023

Performance is a bit worse, but this can save a lot of memory: scroll-tech/halo2#28
Original idea from here: zcash/halo2#427

@Brechtpd Brechtpd converted this from a draft issue Apr 24, 2023
@einar-taiko
Copy link

einar-taiko commented Apr 26, 2023

It looks promising, but AFAICT, it's still WIP: Clippy complains about this statement: if false && condition, so I would suggest we wait until it is finalized.

@Brechtpd
Copy link
Author

Brechtpd commented May 3, 2023

Diff is pretty small, and this will be very important for running the super circuit/root circuit, so would like to get this in sooner than later. Please try it out and see if it works after the FFT, if there aren't any significant issues would like to start using is as soon as possible.

einar-taiko added a commit to einar-taiko/halo2 that referenced this issue Jun 16, 2023
@Brechtpd Brechtpd moved this from 📝 Todo to 🫡 In review in Taiko Project Board Jun 20, 2023
@github-project-automation github-project-automation bot moved this from 🫡 In review to ✅ Done in Taiko Project Board Jun 26, 2023
jonathanpwang pushed a commit to axiom-crypto/halo2 that referenced this issue Oct 26, 2023
jonathanpwang added a commit to axiom-crypto/halo2 that referenced this issue Nov 18, 2023
…17)

* Resolve Prover optimization: memory reduction privacy-scaling-explorations#77 (#6)

* Resolve taikoxyz/zkevm-circuits#77

* Please Clippy

* fix: recursive FFT for lengths not `2^k, 2^extended_k`

Also fixed some issue when multicore feature is not on.

* fix: evaluation (cherry-pick
https://github.com/scroll-tech/halo2-gpu/blob/a3019b047ad7d7119103f4d9df3fd33ce1429f95/halo2_proofs/src/plonk/evaluation.rs)

* fix: `g_to_lagrange` uses inverse fft

* Bump version since this is a breaking change to pk

* chore: update halo2curves version

---------

Co-authored-by: einar-taiko <126954546+einar-taiko@users.noreply.github.com>
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
Status: Done
Development

No branches or pull requests

3 participants