Optimize zkVM Proving by Skipping Unused Keccak Tables #690
Conversation
This reverts commit f63363b.
…le_keccak_optional_proving
sai-deng
force-pushed
the
sai/refactor_optional_table_proving
branch
from
5892a8c
to
27590f8
Compare
|
@Nashtare I’ve refactored the code and merged two tests into one integration test. PTAL. |
zero/src/prover_state/mod.rs
Outdated
| // Default degree of the recursive circuit when a proof is missing from AllProof | ||
| const DEFAULT_CIRCUIT_DEGREE: usize = 8; |
There was a problem hiding this comment.
Why would we want to load anything if a proof is missing? Also this may crash unexpectedly in load_table_circuits if we specify a config with a range (for Keccak for instance) that excludes this value.
There was a problem hiding this comment.
We still need a recursive circuit to generate a dummy proof for the root circuit. Let me refactor the code to optimize this while also addressing the other comment.
sai-deng
force-pushed
the
sai/refactor_optional_table_proving
branch
from
3fd9032
to
9021124
Compare
sai-deng
force-pushed
the
sai/refactor_optional_table_proving
branch
from
667c462
to
557d221
Compare
|
I have changed the code to generate dummy proofs during the initial setup. PTAL again. |
| .as_ref() | ||
| .expect("Missing ctl_zs") | ||
| .clone() | ||
| } else if i == *Table::Keccak { |
There was a problem hiding this comment.
I assume this will get refactored in the generalization to other tables as well?
There was a problem hiding this comment.
Yes. It got refactored in #724
This PR introduces optional Keccak table proving, which optimizes performance by skipping the proving process when the Keccak tables are empty. This change reduces unnecessary computation and requires an update to Plonky2 for full integration.