circuit_definitions
| Crates.io | circuit_definitions |
| lib.rs | circuit_definitions |
| version | 0.140.0-gpu-wrapper.0 |
| source | src |
| created_at | 2024-07-05 12:44:46.237148 |
| updated_at | 2024-07-08 10:24:22.762815 |
| description | ZKsync Era circuits definitions |
| homepage | https://zksync.io/ |
| repository | https://github.com/matter-labs/era-zkevm_test_harness/ |
| max_upload_size | |
| id | 1292719 |
| size | 343,589 |
(ml-ia)
documentation
README
Circuits definitions crate
This crate contains the 'outer layer' for multiple circuits. The concrete circuits code is in era-zkevm_circuits
repository.
Code structure
We have 13 different 'base layer' circuits (for example MainVM, Decomitter), and 3 recursive circuits (Leaf, Node and Scheduler).
Base layer circuits are located in src/base_layer, Recursive circuits are in src/recursion_layer.
We also have 'AUX' circuits: compressors and wrapper, that are run on top of the final Scheduler proof, and they are
located in src/aux_layer.
src/encodings directory contain some helper structs that are used by the test harness (and should match the ones used
in circuits themselves).
Circuit types
We have 12 different circuit types (in witness, you might notice 13, as one circuit (events_dedup_and_sort) is used for both L1 messages and events).
| Circuit name | Location | Description |
|---|---|---|
| Main VM | vm_main.rs | Executes OpCodes from the VM |
| CodeDecommittmentsSorter | sort_code_decommit.rs | Verifies the order of requests for code decommitment (fetching bytecode based on hash). |
| CodeDecommiter | code_decommiter.rs | Unpacks the bytecode matching a given hash into memory. |
| LogDemuxer | log_demux.rs | Splits the queue into 6 output ones (like keccak, sha, storage). |
| KeccakRoundFunction | keccak256_round_functions.rs | Round of the keccak hash |
| Sha256RoundFunction | sha256_round_function.rs | Round of sha256 hash |
| ECRecover | ecrecover.rs | Verifies ECRecover |
| RAMPermutation | ram_permutation.rs | Verifies the correctness of the RAM accesses - looking at the access queue, and checking that correct bytes values were read |
| StorageSorter | storage_sort_dedup.rs | Similar to RAM permutation, but for storage - checking that correct bytes were stored / read. |
| StorageApplication | storage_apply.rs | Verifies the final merkle root and storage diffs based on the data that was written during computation. |
| EventsSorter | events_sort_dedup.rs | Verifies that a given 'unsorted' queue is matching the sorted one, without any repetitions. In this case, used for System Events. |
| L1MessagesSorter | events_sort_dedup.rs | It reuses the circuit above, but this time to sort user generated events (L2 -> L1 messages). |
| L1MessageHasher | linear_hasher.rs | Verifies that linear hash of L1 messages matches the content of the queue. |
3 recursive circuits:
| Circuit name | Location | Description |
|---|---|---|
| Leaf | leaf_layer.rs | Aggregates 32 basic circuits of the same type |
| Node | node_layer.rs | Aggregates 32 leaf (or node) circruits of the same type |
| Scheduler | scheduler.rs | Aggregates 13 nodes (1 from each type) into a final proof |
And 'wrapper'/AUX circuits on top:
| Circuit name | Location | Description |
|---|---|---|
| Compression | compression.rs | Compresses the final scheduler proof |
| Wrapper | wrapper.rs | Wraps the compressed proof into a SNARK to be verifierd on L1. (This is a SNARK circuit) |