feat(avm): plumb execution hints from TS to AVM prover

[dbanks12]

No description provided.

[dbanks12]

• feat(avm): plumb execution hints from TS to AVM prover #6806 : 2 dependent PRs (#6819 , #6848 ) 👈
• master

This stack of pull requests is managed by Graphite. Learn more about stacking.

Join @dbanks12 and the rest of your teammates on Graphite

[AztecBot]

Benchmark results

Metrics with a significant change:

• proof_construction_time_sha256_30_ms (16): 1,752 (+22%)
• protocol_circuit_witness_generation_time_in_ms (base-parity): 1,673 (+89%)
• protocol_circuit_witness_generation_time_in_ms (root-parity): 60.9 (+31%)
• protocol_circuit_proving_time_in_ms (private-kernel-tail): 56,374 (+22%)
• protocol_circuit_proving_time_in_ms (private-kernel-tail-to-public): 102,924 (+21%)
• protocol_circuit_proving_time_in_ms (base-rollup): 78,504 (+20%)
• protocol_circuit_proving_time_in_ms (root-parity): 40,395 (+18%)
• protocol_circuit_proving_time_in_ms (public-kernel-setup): 43,739 (+16%)
• protocol_circuit_proving_time_in_ms (public-kernel-app-logic): 45,350 (+17%)
• protocol_circuit_proving_time_in_ms (public-kernel-teardown): 45,657 (+17%)
• protocol_circuit_proving_time_in_ms (root-rollup): 22,640 (+17%)
• pxe_database_size_in_bytes (3): 7,774 (-16%)

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.

Proof generation

Each column represents the number of threads used in proof generation.

Metric	1 threads	4 threads	16 threads	32 threads	64 threads
proof_construction_time_sha256_ms	5,717 (-3%)	1,547 (-3%)	710 (-1%)	747 (-3%)	776 (-2%)
proof_construction_time_sha256_30_ms	11,740 (-5%)	3,165 (-2%)	⚠️ 1,752 (+22%)	1,581 (+8%)	1,549 (+3%)
proof_construction_time_sha256_100_ms	44,810 (-4%)	11,755 (-3%)	5,422 (-2%)	5,411 (-2%)	5,345 (-2%)
proof_construction_time_poseidon_hash_ms	78.0 (-4%)	34.0 (-3%)	34.0 (-3%)	58.0 (-3%)	89.0 (-1%)
proof_construction_time_poseidon_hash_30_ms	1,518 (-4%)	419 (-3%)	201 (-1%)	228	272 (+1%)
proof_construction_time_poseidon_hash_100_ms	5,754 (-3%)	1,567 (-3%)	720 (-2%)	778 (-2%)	796 (-2%)

L2 block published to L1

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

Metric	4 txs	8 txs
l1_rollup_calldata_size_in_bytes	1,412	1,412
l1_rollup_calldata_gas	9,476	9,456
l1_rollup_execution_gas	608,012	607,992
l2_block_processing_time_in_ms	693 (-1%)	1,292
l2_block_building_time_in_ms	22,605 (+2%)	46,297 (+1%)
l2_block_rollup_simulation_time_in_ms	22,362 (+1%)	45,945 (+1%)
l2_block_public_tx_process_time_in_ms	19,177 (+1%)	42,500 (+1%)

L2 chain processing

Each column represents the number of blocks on the L2 chain where each block has 8 txs.

Metric	3 blocks	5 blocks
node_history_sync_time_in_ms	5,998 (+2%)	8,500 (-1%)
node_database_size_in_bytes	9,490,512	12,746,832
pxe_database_size_in_bytes	⚠️ 7,774 (-16%)	15,127

Circuits stats

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

Circuit	simulation_time_in_ms	witness_generation_time_in_ms	proving_time_in_ms	input_size_in_bytes	output_size_in_bytes	proof_size_in_bytes	num_public_inputs	size_in_gates
private-kernel-init	137 (+1%)	465 (-7%)	12,855 (+5%)	20,634	64,614	89,536	2,731	524,288
private-kernel-inner	407 (-1%)	936 (-7%)	49,210 (+9%)	92,326	64,614	89,536	2,731	2,097,152
private-kernel-tail	583	2,596 (-8%)	⚠️ 56,374 (+22%)	96,545	68,261	11,648	297	2,097,152
base-parity	6.48 (-2%)	⚠️ 1,673 (+89%)	2,683 (-3%)	128	64.0	2,208	2.00	131,072
root-parity	49.5 (-2%)	⚠️ 60.9 (+31%)	⚠️ 40,395 (+18%)	27,100	64.0	2,720	18.0	2,097,152
base-rollup	6,058 (+2%)	2,362	⚠️ 78,504 (+20%)	119,738	756	3,648	47.0	4,194,304
root-rollup	109 (-4%)	80.3 (+29%)	⚠️ 22,640 (+17%)	25,309	620	3,456	41.0	1,048,576
public-kernel-app-logic	573	3,426 (-3%)	⚠️ 45,350 (+17%)	108,073	86,550	116,768	3,582	2,097,152
public-kernel-tail	1,142 (-1%)	23,881 (+1%)	176,313 (+15%)	403,238	7,646	11,648	297	8,388,608
private-kernel-reset-small	589	1,922 (-11%)	45,297 (+11%)	120,737	64,614	89,536	2,731	2,097,152
public-kernel-setup	660 (-4%)	2,718 (+1%)	⚠️ 43,739 (+16%)	108,073	86,550	116,768	3,582	2,097,152
public-kernel-teardown	568 (-3%)	3,468 (-2%)	⚠️ 45,657 (+17%)	108,073	86,550	116,768	3,582	2,097,152
merge-rollup	33.4 (+11%)	N/A	N/A	16,542	756	N/A	N/A	N/A
private-kernel-tail-to-public	N/A	8,274 (-7%)	⚠️ 102,924 (+21%)	N/A	N/A	116,768	3,582	4,194,304

Stats on running time collected for app circuits

Function	input_size_in_bytes	output_size_in_bytes	witness_generation_time_in_ms	proof_size_in_bytes	proving_time_in_ms	size_in_gates	num_public_inputs
ContractClassRegisterer:register	1,344	9,944	456 (-3%)	N/A	N/A	N/A	N/A
ContractInstanceDeployer:deploy	1,408	9,944	41.3 (-2%)	N/A	N/A	N/A	N/A
MultiCallEntrypoint:entrypoint	1,920	9,944	1,767 (-2%)	N/A	N/A	N/A	N/A
SchnorrAccount:constructor	1,312	9,944	1,426 (-3%)	N/A	N/A	N/A	N/A
SchnorrAccount:entrypoint	2,304	9,944	2,809 (-1%)	16,768	53,285 (+6%)	2,097,152	457
Token:privately_mint_private_note	1,280	9,944	1,564 (-7%)	N/A	N/A	N/A	N/A
FPC:fee_entrypoint_public	1,344	9,944	1,016 (-11%)	16,768	10,975 (+9%)	524,288	457
Token:transfer	1,376	9,944	5,305 (-4%)	16,768	52,266 (+10%)	2,097,152	457
Benchmarking:create_note	1,344	9,944	1,394	N/A	N/A	N/A	N/A
SchnorrAccount:spend_private_authwit	1,280	9,944	77.7 (-2%)	N/A	N/A	N/A	N/A
Token:unshield	1,376	9,944	3,911 (-2%)	N/A	N/A	N/A	N/A
FPC:fee_entrypoint_private	1,376	9,944	4,798 (-2%)	N/A	N/A	N/A	N/A

Tree insertion stats

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

Metric	1 leaves	16 leaves	64 leaves	128 leaves	256 leaves	512 leaves	1024 leaves	32 leaves
batch_insert_into_append_only_tree_16_depth_ms	10.4 (-2%)	16.8 (-2%)	N/A	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_16_depth_hash_count	16.7	31.7	N/A	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_16_depth_hash_ms	0.601 (-2%)	0.516 (-2%)	N/A	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_32_depth_ms	N/A	N/A	48.4 (-2%)	75.9 (-1%)	131 (-2%)	247	472 (-2%)	N/A
batch_insert_into_append_only_tree_32_depth_hash_count	N/A	N/A	95.9	159	288	543	1,055	N/A
batch_insert_into_append_only_tree_32_depth_hash_ms	N/A	N/A	0.494 (-2%)	0.468 (-1%)	0.450 (-2%)	0.449	0.441 (-2%)	N/A
batch_insert_into_indexed_tree_20_depth_ms	N/A	N/A	58.5 (-1%)	112 (-1%)	182 (-2%)	352	629 (-1%)	N/A
batch_insert_into_indexed_tree_20_depth_hash_count	N/A	N/A	107	208	355	686	1,263	N/A
batch_insert_into_indexed_tree_20_depth_hash_ms	N/A	N/A	0.504 (-1%)	0.502 (-1%)	0.480 (-1%)	0.481	0.469 (-2%)	N/A
batch_insert_into_indexed_tree_40_depth_ms	N/A	N/A	N/A	N/A	N/A	N/A	N/A	58.8 (-1%)
batch_insert_into_indexed_tree_40_depth_hash_count	N/A	N/A	N/A	N/A	N/A	N/A	N/A	101
batch_insert_into_indexed_tree_40_depth_hash_ms	N/A	N/A	N/A	N/A	N/A	N/A	N/A	0.549 (-1%)

Miscellaneous

Transaction sizes based on how many contract classes are registered in the tx.

Metric	0 registered classes	1 registered classes
tx_size_in_bytes	79,470	665,267

Transaction size based on fee payment method

| Metric | |
| - | |

[dbanks12]

A few caveats with this PR that can be handled in follow-up PRs:

1. This won't work for nested calls because the trace isn't properly separating hints per call
2. The PublicCircuitPublicInputs in the test are empty. I guess we need a function to construct it from an AVM context.
3. We are doing translation between AVM result and PubicExecutionResult in the proving test and that's very ugly

[fcarreiro]

In general it looks ok, given the caveats you mention!

[fcarreiro]

Feels like this PR should include the deserialization of the hints on the CPP side, to be sure that the serde works.

[Maddiaa0]

I can do this in a fast followup

[Maddiaa0]

lgtm