feat: Integrated native ACVM

[PhilWindle]

This PR creates a cli entrypoint for the ACVM simulator, builds it and makes it available for use in e2e tests both locally and on CI.

This native simulator is used to execute sequencer-side protocol circuits, in parallel where possible.

[AztecBot]

Benchmark results

Metrics with a significant change:

• circuit_simulation_time_in_ms (base-rollup): 658 (-53%)
• l2_block_building_time_in_ms (8): 11,276 (-36%)
• l2_block_building_time_in_ms (32): 44,877 (-35%)
• l2_block_building_time_in_ms (64): 89,631 (-37%)
• l2_block_rollup_simulation_time_in_ms (8): 7,322 (-46%)
• l2_block_rollup_simulation_time_in_ms (32): 29,347 (-45%)
• l2_block_rollup_simulation_time_in_ms (64): 58,648 (-46%)
• note_trial_decrypting_time_in_ms (8): 111 (+1144%)

Detailed results

All benchmarks are run on txs on the Benchmarking contract on the repository. Each tx consists of a batch call to create_note and increment_balance, which guarantees that each tx has a private call, a nested private call, a public call, and a nested public call, as well as an emitted private note, an unencrypted log, and public storage read and write.

This benchmark source data is available in JSON format on S3 here.

Values are compared against data from master at commit 06a91169 and shown if the difference exceeds 1%.

L2 block published to L1

Each column represents the number of txs on an L2 block published to L1.

Metric	8 txs	32 txs	64 txs
l1_rollup_calldata_size_in_bytes	5,700	18,884	36,452
l1_rollup_calldata_gas	66,384	239,420	470,216
l1_rollup_execution_gas	194,332	500,726	909,474
l2_block_processing_time_in_ms	1,192 (+1%)	4,476	8,920
note_successful_decrypting_time_in_ms	198 (-2%)	540 (-1%)	1,001
note_trial_decrypting_time_in_ms	⚠️ 111 (+1144%)	61.4 (-15%)	33.9 (-75%)
l2_block_building_time_in_ms	⚠️ 11,276 (-36%)	⚠️ 44,877 (-35%)	⚠️ 89,631 (-37%)
l2_block_rollup_simulation_time_in_ms	⚠️ 7,322 (-46%)	⚠️ 29,347 (-45%)	⚠️ 58,648 (-46%)
l2_block_public_tx_process_time_in_ms	3,925 (-3%)	15,459 (-5%)	30,868 (-7%)

L2 chain processing

Each column represents the number of blocks on the L2 chain where each block has 16 txs.

Metric	5 blocks	10 blocks
node_history_sync_time_in_ms	13,199 (-15%)	27,360 (-3%)
note_history_successful_decrypting_time_in_ms	1,243 (-1%)	2,434 (-1%)
note_history_trial_decrypting_time_in_ms	89.6 (+3%)	139 (-22%)
node_database_size_in_bytes	18,776,144	35,332,176
pxe_database_size_in_bytes	29,923	59,478

Circuits stats

Stats on running time and I/O sizes collected for every circuit run across all benchmarks.

Circuit	circuit_simulation_time_in_ms	circuit_input_size_in_bytes	circuit_output_size_in_bytes
private-kernel-init	255 (-1%)	44,736	28,001
private-kernel-ordering	194 (-2%)	52,625	14,627
base-rollup	⚠️ 658 (-53%)	177,932	997
root-rollup	68.8 (-7%)	4,320	825
private-kernel-inner	323 (-3%)	73,715	28,001
public-kernel-app-logic	173 (-15%)	32,254	25,379
merge-rollup	8.55 (-6%)	2,840	997

Tree insertion stats

The duration to insert a fixed batch of leaves into each tree type.

Metric	1 leaves	2 leaves	8 leaves	16 leaves	32 leaves	64 leaves	128 leaves	512 leaves	1024 leaves	2048 leaves	4096 leaves
batch_insert_into_append_only_tree_16_depth_ms	9.78 (-2%)	10.2 (-7%)	15.2 (+21%)	16.1 (-1%)	21.8 (-1%)	35.5	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_16_depth_hash_count	16.9	17.5	23.0	31.6	47.0	79.0	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_16_depth_hash_ms	0.568 (-2%)	0.569 (-7%)	0.647 (+22%)	0.496 (-1%)	0.458 (-1%)	0.443	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_32_depth_ms	N/A	N/A	N/A	N/A	N/A	45.8 (-1%)	71.8 (-4%)	231	446 (-1%)	867	1,717 (-2%)
batch_insert_into_append_only_tree_32_depth_hash_count	N/A	N/A	N/A	N/A	N/A	96.0	159	543	1,055	2,079	4,127
batch_insert_into_append_only_tree_32_depth_hash_ms	N/A	N/A	N/A	N/A	N/A	0.469 (-1%)	0.443 (-4%)	0.421	0.418	0.413 (-1%)	0.412 (-2%)
batch_insert_into_indexed_tree_20_depth_ms	N/A	N/A	N/A	N/A	N/A	54.0 (-4%)	107	338 (+1%)	659 (-1%)	1,307	2,614
batch_insert_into_indexed_tree_20_depth_hash_count	N/A	N/A	N/A	N/A	N/A	104	207	691	1,363	2,707	5,395
batch_insert_into_indexed_tree_20_depth_hash_ms	N/A	N/A	N/A	N/A	N/A	0.481 (-3%)	0.481	0.461 (+1%)	0.454	0.454	0.453
batch_insert_into_indexed_tree_40_depth_ms	N/A	N/A	N/A	N/A	61.1 (-1%)	N/A	N/A	N/A	N/A	N/A	N/A
batch_insert_into_indexed_tree_40_depth_hash_count	N/A	N/A	N/A	N/A	109	N/A	N/A	N/A	N/A	N/A	N/A
batch_insert_into_indexed_tree_40_depth_hash_ms	N/A	N/A	N/A	N/A	0.535 (-1%)	N/A	N/A	N/A	N/A	N/A	N/A

Miscellaneous

Transaction sizes based on how many contracts are deployed in the tx.

Metric	0 deployed contracts
tx_size_in_bytes	19,179

Transaction processing duration by data writes.

Metric	0 new note hashes	1 new note hashes
tx_pxe_processing_time_ms	2,712 (+2%)	1,379 (-2%)

Metric	0 public data writes	1 public data writes
tx_sequencer_processing_time_ms	0.0309 (+3%)	480 (-3%)

[socket-security]

No dependency changes detected. Learn more about Socket for GitHub ↗︎

👍 No dependency changes detected in pull request

[alexghr]

Code looks good, but I'm surprised to see some of the circuits take longer to complete in the bench e2e test (according to the comment). Assuming that the bench uses the native simulator, could the slowdown come from mapping data and writing/reading to disk?

Later edit: looks like switching back to WASM for the smaller circuits completely fixed the perf regressions 🥳

[alexghr]

🔥