chore: charge for more l2 gas costs

[just-mitch]

charge l2 gas for:

• avm startup
• note hashes in private
• nullifiers in private
• logs in private

[AztecBot]

Docs Preview

Hey there! 👋 You can check your preview at https://6685d43961c16b1a26db6469--aztec-docs-dev.netlify.app

[AztecBot]

Benchmark results

Metrics with a significant change:

• avm_simulation_time_ms (Token:mint_public): 532 (+683%)
• avm_simulation_time_ms (Token:assert_minter_and_mint): 59.3 (-68%)
• avm_simulation_time_ms (Token:transfer_public): 58.9 (+44%)

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	5,738	1,561 (+1%)	708	750 (-1%)	777 (-1%)
proof_construction_time_sha256_30_ms	11,935 (+1%)	3,179 (+1%)	1,408 (-1%)	1,435 (+1%)	1,473 (+1%)
proof_construction_time_sha256_100_ms	44,113 (+1%)	11,839 (+1%)	5,451	5,418 (-1%)	5,383
proof_construction_time_poseidon_hash_ms	78.0	35.0 (+3%)	34.0	58.0 (-2%)	87.0 (-2%)
proof_construction_time_poseidon_hash_30_ms	1,527 (+1%)	419 (+1%)	201 (+1%)	228 (+1%)	269 (+2%)
proof_construction_time_poseidon_hash_100_ms	5,793 (+1%)	1,580 (+1%)	724	772 (-2%)	799 (+1%)

L2 block published to L1

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

Metric	4 txs	8 txs	16 txs
l1_rollup_calldata_size_in_bytes	1,412	1,412	1,412
l1_rollup_calldata_gas	9,476	9,460	9,464
l1_rollup_execution_gas	611,215	611,350	611,505
l2_block_processing_time_in_ms	761	1,411	2,692 (-1%)
l2_block_building_time_in_ms	20,911 (-1%)	41,801	81,635 (-1%)
l2_block_rollup_simulation_time_in_ms	20,911 (-1%)	41,801	81,634 (-1%)
l2_block_public_tx_process_time_in_ms	17,884 (-1%)	38,579	78,474 (-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	7,002	9,802 (-1%)
node_database_size_in_bytes	12,283,984	16,162,896
pxe_database_size_in_bytes	16,254	26,813

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	102 (-1%)	388 (+2%)	12,725 (+4%)	19,482	54,134	73,920	2,243	524,288
private-kernel-inner	307	787 (+4%)	50,383	80,694	54,134	73,920	2,243	2,097,152
private-kernel-tail	1,094 (+1%)	2,705 (+7%)	45,854 (-10%)	61,457	62,057	14,912	399	2,097,152
base-parity	6.12 (-2%)	1,541 (-1%)	2,633 (+1%)	128	64.0	2,208	2.00	131,072
root-parity	49.3 (+2%)	67.3 (+5%)	42,431 (+1%)	27,100	64.0	2,720	18.0	2,097,152
base-rollup	6,582	4,881	89,695	170,330	728	3,648	47.0	4,194,304
root-rollup	111 (+1%)	87.9 (+10%)	22,521 (-7%)	25,253	620	3,456	41.0	1,048,576
public-kernel-setup	544 (+1%)	2,374	42,030 (-4%)	102,121	80,278	106,912	3,274	2,097,152
public-kernel-app-logic	500	3,289 (-1%)	43,005 (+1%)	102,121	80,278	106,912	3,274	2,097,152
public-kernel-tail	1,146	23,785 (-10%)	176,599 (+1%)	399,014	10,014	14,912	399	8,388,608
private-kernel-reset-small	467	1,074 (-3%)	27,087 (-11%)	109,233	54,134	73,920	2,243	1,048,576
public-kernel-teardown	493 (-6%)	3,339 (+1%)	42,771 (-4%)	102,121	80,278	106,912	3,274	2,097,152
merge-rollup	29.3	N/A	N/A	16,486	728	N/A	N/A	N/A
private-kernel-tail-to-public	N/A	8,393 (-4%)	50,025 (-5%)	N/A	N/A	106,912	3,274	2,097,152

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	8,792	407 (+1%)	N/A	N/A	N/A	N/A
ContractInstanceDeployer:deploy	1,408	8,792	38.7 (+1%)	N/A	N/A	N/A	N/A
MultiCallEntrypoint:entrypoint	1,920	8,792	1,189	N/A	N/A	N/A	N/A
GasToken:deploy	1,376	8,792	894 (+2%)	N/A	N/A	N/A	N/A
SchnorrAccount:constructor	1,312	8,792	488 (-1%)	N/A	N/A	N/A	N/A
SchnorrAccount:entrypoint	2,304	8,792	1,619	14,720	53,320	2,097,152	393
Token:privately_mint_private_note	1,280	8,792	617	N/A	N/A	N/A	N/A
FPC:fee_entrypoint_public	1,344	8,792	267	14,720	12,257 (+4%)	524,288	393
Token:transfer	1,312	8,792	1,768	14,720	12,538 (-6%)	524,288	393
AuthRegistry:set_authorized (avm)	19,226	N/A	N/A	91,264	1,350 (+8%)	N/A	N/A
FPC:prepare_fee (avm)	26,668	N/A	N/A	91,328	2,773 (+2%)	N/A	N/A
Token:transfer_public (avm)	42,918	N/A	N/A	91,328	3,833	N/A	N/A
AuthRegistry:consume (avm)	33,104	N/A	N/A	91,264	2,950 (+9%)	N/A	N/A
FPC:pay_refund (avm)	36,833	N/A	N/A	91,296	23,114	N/A	N/A
Benchmarking:create_note	1,344	8,792	480	N/A	N/A	N/A	N/A
SchnorrAccount:verify_private_authwit	1,280	8,792	71.3 (-10%)	N/A	N/A	N/A	N/A
Token:unshield	1,376	8,792	1,520 (-1%)	N/A	N/A	N/A	N/A
FPC:fee_entrypoint_private	1,376	8,792	2,103 (-1%)	N/A	N/A	N/A	N/A

AVM Simulation

Time to simulate various public functions in the AVM.

Function	time_ms	bytecode_size_in_bytes
GasToken:_increase_public_balance	69.3 (+3%)	13,790
GasToken:set_portal	14.2 (-19%)	3,339
Token:constructor	92.6 (+1%)	23,692
FPC:constructor	61.0 (-3%)	13,592
GasToken:mint_public	54.1 (+7%)	10,158
Token:mint_public	⚠️ 532 (+683%)	19,034
Token:assert_minter_and_mint	⚠️ 59.3 (-68%)	12,925
AuthRegistry:set_authorized	35.9 (+14%)	7,812
FPC:prepare_fee	107 (-1%)	15,062
Token:transfer_public	⚠️ 58.9 (+44%)	31,218
FPC:pay_refund	128 (-13%)	25,260
Benchmarking:increment_balance	2,198	15,267
Token:_increase_public_balance	56.1	15,006
FPC:pay_refund_with_shielded_rebate	116 (-1%)	26,347

Public DB Access

Time to access various public DBs.

Function	time_ms
get-nullifier-index	0.156 (-5%)

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
batch_insert_into_append_only_tree_16_depth_ms	10.5 (+1%)	16.8 (+1%)	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_16_depth_hash_count	16.8	31.7	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_16_depth_hash_ms	0.608 (+1%)	0.517 (+1%)	N/A	N/A	N/A	N/A	N/A
batch_insert_into_append_only_tree_32_depth_ms	N/A	N/A	48.3 (+1%)	75.8	131 (-1%)	245	470 (-1%)
batch_insert_into_append_only_tree_32_depth_hash_count	N/A	N/A	95.9	159	287	543	1,055
batch_insert_into_append_only_tree_32_depth_hash_ms	N/A	N/A	0.494 (+1%)	0.466	0.450 (-1%)	0.444	0.439
batch_insert_into_indexed_tree_20_depth_ms	N/A	N/A	59.3	112	183 (-1%)	352	694
batch_insert_into_indexed_tree_20_depth_hash_count	N/A	N/A	109	207	355	691	1,363
batch_insert_into_indexed_tree_20_depth_hash_ms	N/A	N/A	0.501	0.501	0.484	0.477	0.476
batch_insert_into_indexed_tree_40_depth_ms	N/A	N/A	72.7	N/A	N/A	N/A	N/A
batch_insert_into_indexed_tree_40_depth_hash_count	N/A	N/A	133	N/A	N/A	N/A	N/A
batch_insert_into_indexed_tree_40_depth_hash_ms	N/A	N/A	0.518	N/A	N/A	N/A	N/A

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	74,057	667,850

Transaction size based on fee payment method

| Metric | |
| - | |

[fcarreiro]

Don't you want this for both branches?

[just-mitch]

Yes, for sure. My mistake.

[fcarreiro]

Are you aware making this change here in the machine state will also charge the cost for each external call (even each nested)? (even down the road when we do 1 enqueued call per AVM circuit)

[just-mitch]

Yes, that was the intent, because as it stands we need to generate a public kernel proof for each external/nested call right? And that is what this startup cost is primarily intended to cover.

[fcarreiro]

As I mentioned in the chat I think this would break (or at least misalign) the AVM circuit. (@dbanks12 correct me if I'm wrong).

Connecting to my other comment about doing this every enqueued call vs every nested call/avm circuit: did you consider making this gas consumption at the abstract phase manager or publicexecutor level?

Note that, on the one hand, that probably lets you just do the circuit equivalent in one of the public kernels. However, doing it in the abstract phase manager/public executor will not let you do this for every nested call (/current avm circuit execution).

[just-mitch]

I wanted it to happen in the AVM so that the public kernel wouldn't need to start doing its own "out of gas" checks, since it presently just accepts the result of the AVM. I'm happy to discuss alternatives though!

[dbanks12]

For nested calls, I think this constant is unnecessary as it can just be considered part of the CALL & STATICCALL instruction cost.

This base cost might make sense for top-level enqueued calls since they AVM itself doesn't trigger a top-level call via an instruction.

Could that base gas instead be charged at the time a public call is enqueued in private land?

At the offsite we made substantial simplifications to how the AVM computes gas per-instruction because the circuit isn't ready for more complex gas computations. And we are now at the stage where we want the simulator and circuit to align. For now I believe we just a have a constant lookup table for gas (opcode -> const gas cost).

[just-mitch]

@dbanks12 Ah yes of course!

Then it will follow the same pattern as notes/nullifiers where you get charged L2 gas in private, and we bake in additional L2 gas (32/64) when we EMITNOTEHASH/EMITNULLIFIER.

[fcarreiro]

Can this potentially make the gas left negative or undeflow?

[just-mitch]

Yes! Fixed this.

[just-mitch]

Okay @dbanks12 @fcarreiro all AVM Simulator stuff is removed 👍 .

[MirandaWood]

Looks good for the protocol circuits!