feat: benchmark private proving

[alexghr]

This PR adds benchmarks for the private kernel running with real BB proofs. This PR also tracks these stats for individual function calls.

[alexghr]

• feat: benchmark private proving #6409 👈
• master

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

Join @alexghr and the rest of your teammates on Graphite

[PhilWindle]

A few things but almost there

[PhilWindle]

Could we make the inputSize simply compressedBincodedWitness.length?

[PhilWindle]

Same here

[PhilWindle]

Probably better named protocol-circuit-name.

[PhilWindle]

I'd like to capture the length of the serialised inputs struct if we can.

[PhilWindle]

Can you have 2 columns. One of which is the actual proof size, and the other the number of public inputs (available on VerificationKeyData)?

[AztecBot]

Benchmark results

Metrics with a significant change:

• tx_pxe_processing_time_ms (0): 28,137 (+849%)
• tx_pxe_processing_time_ms (2): 91,844 (+1128%)

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.

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	740	740	740
l1_rollup_calldata_gas	6,680	6,680	6,680
l1_rollup_execution_gas	586,865	586,865	586,865
l2_block_processing_time_in_ms	1,603 (-1%)	6,013 (-1%)	11,965 (-1%)
l2_block_building_time_in_ms	27,254	107,094	214,282
l2_block_rollup_simulation_time_in_ms	27,074	106,454	213,029
l2_block_public_tx_process_time_in_ms	11,182	44,060	88,100 (-1%)

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	18,264 (+1%)	33,578
node_database_size_in_bytes	22,859,856	39,448,656
pxe_database_size_in_bytes	29,868	59,425

Circuits stats

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

Circuit	protocol_circuit_simulation_time_in_ms	protocol_circuit_witness_generation_time_in_ms	protocol_circuit_proving_time_in_ms	protocol_circuit_input_size_in_bytes	protocol_circuit_output_size_in_bytes	protocol_circuit_proof_size_in_bytes	protocol_circuit_num_public_inputs	protocol_circuit_size_in_gates
private-kernel-init	144	2,601	20,999	19,396	46,118	62,400	1,883	1,048,576
private-kernel-inner	421	3,733	38,209	72,583	46,118	62,400	1,883	2,097,152
private-kernel-reset	682	4,430	45,198	219,992	46,118	62,400	1,883	2,097,152
private-kernel-tail	448	4,495	37,331	70,968	64,320	10,656	266	2,097,152
base-parity	8.16	109	2,746	128	64.0	2,208	2.00	131,072
root-parity	51.8	130	40,041	27,064	64.0	2,720	18.0	2,097,152
base-rollup	725	5,338	42,954	111,126	925	3,104	30.0	2,097,152
root-rollup	72.6	173	8,221	11,454	789	3,424	40.0	524,288
public-kernel-app-logic	156	250	777	78,387	68,775	90,432	2,759	4,096
public-kernel-tail	540	597	847	184,419	7,723	10,144	250	512
public-kernel-setup	129	333	815	110,522	68,775	90,432	2,759	4,096
public-kernel-teardown	130	347	825	114,417	68,775	90,432	2,759	4,096
merge-rollup	6.66	88.0	1,542	2,696	925	3,104	30.0	65,536
private-kernel-tail-to-public	N/A	9,075	43,207	N/A	N/A	90,944	2,775	2,097,152

Stats on running time collected for app circuits

Function	app_circuit_proof_size_in_bytes	app_circuit_proving_time_in_ms	app_circuit_size_in_gates	app_circuit_num_public_inputs
SchnorrAccount:entrypoint	13,600	45,814	2,097,152	358
Test:emit_nullifier	13,600	2,326	65,536	358
FPC:fee_entrypoint_public	13,600	13,488	524,288	358
Token:transfer	13,600	43,392	2,097,152	358
FPC:fee_entrypoint_private	13,600	11,182	524,288	358
Token:unshield	13,600	33,679	2,097,152	358
SchnorrAccount:spend_private_authwit	13,600	2,617	131,072	358

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	512 leaves	1024 leaves	2048 leaves	4096 leaves	32 leaves
batch_insert_into_append_only_tree_16_depth_ms	13.1	21.4	N/A	N/A	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_16_depth_hash_count	16.7	31.8	N/A	N/A	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_16_depth_hash_ms	0.766	0.658	N/A	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	61.7 (-1%)	97.2 (-2%)	317 (-1%)	607 (+1%)	1,194 (-1%)	2,360 (-1%)	N/A
batch_insert_into_append_only_tree_32_depth_hash_count	N/A	N/A	95.9	159	543	1,055	2,079	4,127	N/A
batch_insert_into_append_only_tree_32_depth_hash_ms	N/A	N/A	0.632 (-1%)	0.602 (-2%)	0.577 (-1%)	0.568 (+1%)	0.567	0.565 (-1%)	N/A
batch_insert_into_indexed_tree_20_depth_ms	N/A	N/A	73.2	143	453	889	1,765 (-1%)	3,522 (-1%)	N/A
batch_insert_into_indexed_tree_20_depth_hash_count	N/A	N/A	106	208	692	1,363	2,707	5,395	N/A
batch_insert_into_indexed_tree_20_depth_hash_ms	N/A	N/A	0.641	0.643	0.618	0.613	0.613 (-1%)	0.614 (-1%)	N/A
batch_insert_into_indexed_tree_40_depth_ms	N/A	N/A	N/A	N/A	N/A	N/A	N/A	N/A	80.4
batch_insert_into_indexed_tree_40_depth_hash_count	N/A	N/A	N/A	N/A	N/A	N/A	N/A	N/A	108
batch_insert_into_indexed_tree_40_depth_hash_ms	N/A	N/A	N/A	N/A	N/A	N/A	N/A	N/A	0.708

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	68,448 (+1%)	1,052,029

Transaction size based on fee payment method

| Metric | |
| - | |

Transaction processing duration by data writes.

Metric	0 new note hashes	1 new note hashes	2 new note hashes
tx_pxe_processing_time_ms	⚠️ 28,137 (+849%)	3,074 (+1%)	⚠️ 91,844 (+1128%)

Metric	0 public data writes	1 public data writes	2 public data writes	4 public data writes	8 public data writes
tx_sequencer_processing_time_ms	806 (-3%)	1,118	1,267 (+7%)	2,012 (-9%)	1,360