Skip to content

Conversation

@hero78119
Copy link
Collaborator

@hero78119 hero78119 commented Jan 20, 2026

To close #1232

Move ecc quark major mles of septic curve and sumcheck to GPU. selector remain in cpu as its overhead are negligible, due to the num rows in ecc quark are relatively small, .e.g ~2^21.

benchmark on larger block 24169100

23817600 block are relative small where only 14 shards, thus we can only see ~2s e2e different

24169100 with 24 shards and each have shard proofs

Stage / Component Before Time After Time
reth-block (block 24169100) 249 s 240 s
app.prove 169 s 160 s
prove_ec_sum_quark (shard id 0) 640 ms 49.2 ms
prove_ec_sum_quark (shard id 1) 682 ms 49.6 ms
prove_ec_sum_quark (shard id 2) 650 ms 50.6 ms
prove_ec_sum_quark (shard id 3) 355 ms 28.3 ms
...

@hero78119 hero78119 force-pushed the feat/ecc_quark_gpu_prover branch from e7d86c4 to df2731c Compare January 20, 2026 08:21
last_layer
})
.collect::<Vec<_>>()
} else if lk_denominator_last_layer.is_empty() {
Copy link
Collaborator Author

Choose a reason for hiding this comment

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

No logic change, just make clippy happy

@hero78119 hero78119 changed the title Feat/ecc quark gpu prover ecc quark gpu prover Jan 20, 2026
@hero78119 hero78119 force-pushed the feat/ecc_quark_gpu_prover branch 4 times, most recently from 538568f to ce129c6 Compare January 20, 2026 11:52
@hero78119 hero78119 force-pushed the feat/ecc_quark_gpu_prover branch from ce129c6 to 0dc3a37 Compare January 20, 2026 12:26
@hero78119 hero78119 added this pull request to the merge queue Jan 21, 2026
Merged via the queue into master with commit 73f28ef Jan 21, 2026
4 checks passed
@hero78119 hero78119 deleted the feat/ecc_quark_gpu_prover branch January 21, 2026 09:10
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.

EC quark prover in GPU

3 participants