8000 feat: Use SP1 prover directly by wwared · Pull Request #413 · lurk-lab/lurk · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

feat: Use SP1 prover directly #413

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 8 commits into from
Mar 18, 2025
Merged

feat: Use SP1 prover directly #413

merged 8 commits into from
Mar 18, 2025

Conversation

wwared
Copy link
Contributor
@wwared wwared commented Jan 18, 2025

This PR entirely removes all Sphinx dependencies and replaces them with their respective SP1 counterparts, using the latest version published to crates.io.

Because of performance-oriented changes in SP1 v4, the lookup argument for cross-shard communication was redesigned, and our current implementation now only works in a single shard temporarily. A workaround is possible, but was not implemented to minimize complexity (see #437). The impact is only for long computations, and those already had other underlying issues, that this PR does not aim to fix.

As a result to this, there might be a regression in our test runtimes or benchmarks, but everything else should still be mostly the same.

The plonky3 crates are using released versions, except for p3-poseidon2, which is using a slightly patched branch in https://github.com/lurk-lab/p3/tree/sp1-v4

@wwared wwared changed the title Feat: Use SP1 prover directly feat: Use SP1 prover directly Jan 18, 2025
wwared added 5 commits March 13, 2025 12:07
Removes the PrimeField32 bound where unnecessary
This impacts mainly long computations with a lot of sharding (see
lair_shard_test) and large matrices, or large lair programs
After SP1 v4, the global lookup argument for cross-shard chip
communication was drastically redesigned, which means our currently
implementation is temporarily broken. See #437. The proper solution will
require a proper design for sharded Lurk computation.

This commit also removes all of the now-unnecessary dummy chips, since
only single-shard computations can be proven right now.
@wwared wwared marked this pull request as ready for review March 13, 2025 15:25
@wwared wwared enabled auto-merge March 13, 2025 18:12
Copy link
Contributor
@winston-h-zhang winston-h-zhang left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@wwared There seem to be minimal deep changes other than removing sphinx stuff and using Arc. This looks good to me!

@wwared wwared added this pull request to the merge queue Mar 18, 2025
Merged via the queue into main with commit b8236c1 Mar 18, 2025
2 checks passed
@wwared wwared deleted the rebase-sp1 branch March 18, 2025 00:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants
0