Progressively compute target stake

[tuyennhv]

Motivation

Research into computing target stake progressively to compute participation through an epoch instead of at the end of epoch.

Description

• Add currentTargetUnslashedBalanceIncrements, previoustTargetUnslashedBalanceIncrements and compute it per block processing
• Add assertion in beforeEpochProcess, by default do the assertion in spec tests
• Also add cli flag to do the assertion, which allow us to sync from genesis and test

[github-actions]

Performance Report

✔️ no performance regression detected

Full benchmark results

Benchmark suite	Current: f3df2a4	Previous: 7478093	Ratio
getPubkeys - index2pubkey - req 1000 vs - 250000 vc	2.2781 ms/op	2.6862 ms/op	0.85
getPubkeys - validatorsArr - req 1000 vs - 250000 vc	79.828 us/op	90.940 us/op	0.88
BLS verify - blst-native	2.6285 ms/op	2.1992 ms/op	1.20
BLS verifyMultipleSignatures 3 - blst-native	5.0633 ms/op	4.5196 ms/op	1.12
BLS verifyMultipleSignatures 8 - blst-native	11.329 ms/op	10.352 ms/op	1.09
BLS verifyMultipleSignatures 32 - blst-native	41.220 ms/op	35.863 ms/op	1.15
BLS aggregatePubkeys 32 - blst-native	56.478 us/op	45.232 us/op	1.25
BLS aggregatePubkeys 128 - blst-native	210.39 us/op	173.40 us/op	1.21
getAttestationsForBlock	53.356 ms/op	54.638 ms/op	0.98
isKnown best case - 1 super set check	481.00 ns/op	494.00 ns/op	0.97
isKnown normal case - 2 super set checks	485.00 ns/op	463.00 ns/op	1.05
isKnown worse case - 16 super set checks	483.00 ns/op	490.00 ns/op	0.99
CheckpointStateCache - add get delete	13.432 us/op	12.246 us/op	1.10
validate gossip signedAggregateAndProof - struct	6.2344 ms/op	4.6693 ms/op	1.34
validate gossip attestation - struct	2.9438 ms/op	2.2562 ms/op	1.30
altair verifyImport mainnet_s3766816:31	14.741 s/op	7.3320 s/op	2.01
pickEth1Vote - no votes	2.2619 ms/op	2.5005 ms/op	0.90
pickEth1Vote - max votes	25.866 ms/op	31.115 ms/op	0.83
pickEth1Vote - Eth1Data hashTreeRoot value x2048	13.710 ms/op	15.026 ms/op	0.91
pickEth1Vote - Eth1Data hashTreeRoot tree x2048	23.812 ms/op	24.468 ms/op	0.97
pickEth1Vote - Eth1Data fastSerialize value x2048	1.7402 ms/op	1.6895 ms/op	1.03
pickEth1Vote - Eth1Data fastSerialize tree x2048	18.865 ms/op	20.194 ms/op	0.93
bytes32 toHexString	1.1790 us/op	1.3390 us/op	0.88
bytes32 Buffer.toString(hex)	724.00 ns/op	823.00 ns/op	0.88
bytes32 Buffer.toString(hex) from Uint8Array	984.00 ns/op	1.0360 us/op	0.95
bytes32 Buffer.toString(hex) + 0x	778.00 ns/op	715.00 ns/op	1.09
Object access 1 prop	0.40900 ns/op	0.36200 ns/op	1.13
Map access 1 prop	0.31600 ns/op	0.32000 ns/op	0.99
Object get x1000	14.834 ns/op	19.549 ns/op	0.76
Map get x1000	1.1420 ns/op	1.0110 ns/op	1.13
Object set x1000	109.03 ns/op	123.92 ns/op	0.88
Map set x1000	78.682 ns/op	68.263 ns/op	1.15
Return object 10000 times	0.39690 ns/op	0.39420 ns/op	1.01
Throw Error 10000 times	6.5715 us/op	6.3649 us/op	1.03
enrSubnets - fastDeserialize 64 bits	3.0990 us/op	2.7730 us/op	1.12
enrSubnets - ssz BitVector 64 bits	1.1380 us/op	737.00 ns/op	1.54
enrSubnets - fastDeserialize 4 bits	615.00 ns/op	428.00 ns/op	1.44
enrSubnets - ssz BitVector 4 bits	1.0690 us/op	740.00 ns/op	1.44
prioritizePeers score -10:0 att 32-0.1 sync 2-0	102.36 us/op	96.472 us/op	1.06
prioritizePeers score 0:0 att 32-0.25 sync 2-0.25	140.86 us/op	148.78 us/op	0.95
prioritizePeers score 0:0 att 32-0.5 sync 2-0.5	251.65 us/op	246.26 us/op	1.02
prioritizePeers score 0:0 att 64-0.75 sync 4-0.75	544.36 us/op	411.74 us/op	1.32
prioritizePeers score 0:0 att 64-1 sync 4-1	587.70 us/op	499.38 us/op	1.18
RateTracker 1000000 limit, 1 obj count per request	201.38 ns/op	217.25 ns/op	0.93
RateTracker 1000000 limit, 2 obj count per request	151.72 ns/op	172.79 ns/op	0.88
RateTracker 1000000 limit, 4 obj count per request	119.44 ns/op	151.05 ns/op	0.79
RateTracker 1000000 limit, 8 obj count per request	113.09 ns/op	137.09 ns/op	0.82
RateTracker with prune	4.9490 us/op	4.9390 us/op	1.00
array of 16000 items push then shift	4.6075 us/op	3.4446 us/op	1.34
LinkedList of 16000 items push then shift	24.844 ns/op	26.447 ns/op	0.94
array of 16000 items push then pop	233.72 ns/op	271.49 ns/op	0.86
LinkedList of 16000 items push then pop	21.551 ns/op	22.488 ns/op	0.96
array of 24000 items push then shift	7.3977 us/op	5.0316 us/op	1.47
LinkedList of 24000 items push then shift	25.364 ns/op	27.577 ns/op	0.92
array of 24000 items push then pop	186.51 ns/op	217.16 ns/op	0.86
LinkedList of 24000 items push then pop	21.259 ns/op	23.145 ns/op	0.92
intersect bitArray bitLen 8	11.477 ns/op	12.879 ns/op	0.89
intersect array and set length 8	178.91 ns/op	170.74 ns/op	1.05
intersect bitArray bitLen 128	66.101 ns/op	69.535 ns/op	0.95
intersect array and set length 128	2.3962 us/op	2.4601 us/op	0.97
pass gossip attestations to forkchoice per slot	3.1768 ms/op	7.0036 ms/op	0.45
computeDeltas	3.2375 ms/op	3.9859 ms/op	0.81
computeProposerBoostScoreFromBalances	822.80 us/op	1.0272 ms/op	0.80
altair processAttestation - 250000 vs - 7PWei normalcase	5.0529 ms/op	4.4347 ms/op	1.14
altair processAttestation - 250000 vs - 7PWei worstcase	7.2443 ms/op	6.3853 ms/op	1.13
altair processAttestation - setStatus - 1/6 committees join	245.09 us/op	220.93 us/op	1.11
altair processAttestation - setStatus - 1/3 committees join	421.13 us/op	441.75 us/op	0.95
altair processAttestation - setStatus - 1/2 committees join	635.43 us/op	647.11 us/op	0.98
altair processAttestation - setStatus - 2/3 committees join	817.34 us/op	804.94 us/op	1.02
altair processAttestation - setStatus - 4/5 committees join	1.1739 ms/op	1.0894 ms/op	1.08
altair processAttestation - setStatus - 100% committees join	1.4581 ms/op	1.3109 ms/op	1.11
altair processBlock - 250000 vs - 7PWei normalcase	27.721 ms/op	30.358 ms/op	0.91
altair processBlock - 250000 vs - 7PWei normalcase hashState	38.895 ms/op	40.843 ms/op	0.95
altair processBlock - 250000 vs - 7PWei worstcase	100.88 ms/op	92.763 ms/op	1.09
altair processBlock - 250000 vs - 7PWei worstcase hashState	109.79 ms/op	105.05 ms/op	1.05
phase0 processBlock - 250000 vs - 7PWei normalcase	4.6757 ms/op	3.8630 ms/op	1.21
phase0 processBlock - 250000 vs - 7PWei worstcase	60.186 ms/op	50.226 ms/op	1.20
altair processEth1Data - 250000 vs - 7PWei normalcase	999.33 us/op	1.0532 ms/op	0.95
Tree 40 250000 create	990.06 ms/op	882.23 ms/op	1.12
Tree 40 250000 get(125000)	309.21 ns/op	344.27 ns/op	0.90
Tree 40 250000 set(125000)	3.6894 us/op	3.0415 us/op	1.21
Tree 40 250000 toArray()	33.562 ms/op	37.520 ms/op	0.89
Tree 40 250000 iterate all - toArray() + loop	36.774 ms/op	37.732 ms/op	0.97
Tree 40 250000 iterate all - get(i)	123.10 ms/op	132.82 ms/op	0.93
MutableVector 250000 create	14.515 ms/op	18.869 ms/op	0.77
MutableVector 250000 get(125000)	13.148 ns/op	15.451 ns/op	0.85
MutableVector 250000 set(125000)	689.61 ns/op	663.20 ns/op	1.04
MutableVector 250000 toArray()	6.5358 ms/op	7.5228 ms/op	0.87
MutableVector 250000 iterate all - toArray() + loop	6.2910 ms/op	7.8403 ms/op	0.80
MutableVector 250000 iterate all - get(i)	3.4211 ms/op	3.7611 ms/op	0.91
Array 250000 create	5.7923 ms/op	7.0702 ms/op	0.82
Array 250000 clone - spread	3.0944 ms/op	4.5403 ms/op	0.68
Array 250000 get(125000)	1.3110 ns/op	1.9940 ns/op	0.66
Array 250000 set(125000)	1.4260 ns/op	1.9620 ns/op	0.73
Array 250000 iterate all - loop	144.86 us/op	198.30 us/op	0.73
effectiveBalanceIncrements clone Uint8Array 300000	275.27 us/op	89.045 us/op	3.09
effectiveBalanceIncrements clone MutableVector 300000	711.00 ns/op	756.00 ns/op	0.94
effectiveBalanceIncrements rw all Uint8Array 300000	298.11 us/op	277.49 us/op	1.07
effectiveBalanceIncrements rw all MutableVector 300000	204.13 ms/op	168.69 ms/op	1.21
phase0 afterProcessEpoch - 250000 vs - 7PWei	196.79 ms/op	190.15 ms/op	1.03
phase0 beforeProcessEpoch - 250000 vs - 7PWei	64.663 ms/op	73.638 ms/op	0.88
altair processEpoch - mainnet_e81889	632.92 ms/op	652.98 ms/op	0.97
mainnet_e81889 - altair beforeProcessEpoch	98.768 ms/op	184.87 ms/op	0.53
mainnet_e81889 - altair processJustificationAndFinalization	61.327 us/op	29.798 us/op	2.06
mainnet_e81889 - altair processInactivityUpdates	10.774 ms/op	12.425 ms/op	0.87
mainnet_e81889 - altair processRewardsAndPenalties	158.57 ms/op	108.21 ms/op	1.47
mainnet_e81889 - altair processRegistryUpdates	15.654 us/op	7.8650 us/op	1.99
mainnet_e81889 - altair processSlashings	3.7390 us/op	2.1020 us/op	1.78
mainnet_e81889 - altair processEth1DataReset	4.2210 us/op	2.3290 us/op	1.81
mainnet_e81889 - altair processEffectiveBalanceUpdates	2.9680 ms/op	2.7101 ms/op	1.10
mainnet_e81889 - altair processSlashingsReset	29.091 us/op	13.826 us/op	2.10
mainnet_e81889 - altair processRandaoMixesReset	28.646 us/op	11.463 us/op	2.50
mainnet_e81889 - altair processHistoricalRootsUpdate	3.8240 us/op	2.4290 us/op	1.57
mainnet_e81889 - altair processParticipationFlagUpdates	17.326 us/op	7.1450 us/op	2.42
mainnet_e81889 - altair processSyncCommitteeUpdates	3.8700 us/op	1.7250 us/op	2.24
mainnet_e81889 - altair afterProcessEpoch	201.14 ms/op	220.84 ms/op	0.91
phase0 processEpoch - mainnet_e58758	618.44 ms/op	598.58 ms/op	1.03
mainnet_e58758 - phase0 beforeProcessEpoch	279.82 ms/op	262.21 ms/op	1.07
mainnet_e58758 - phase0 processJustificationAndFinalization	59.750 us/op	32.127 us/op	1.86
mainnet_e58758 - phase0 processRewardsAndPenalties	78.891 ms/op	152.78 ms/op	0.52
mainnet_e58758 - phase0 processRegistryUpdates	39.753 us/op	23.801 us/op	1.67
mainnet_e58758 - phase0 processSlashings	3.9500 us/op	2.8430 us/op	1.39
mainnet_e58758 - phase0 processEth1DataReset	4.8690 us/op	2.4450 us/op	1.99
mainnet_e58758 - phase0 processEffectiveBalanceUpdates	2.2840 ms/op	2.6006 ms/op	0.88
mainnet_e58758 - phase0 processSlashingsReset	20.271 us/op	12.009 us/op	1.69
mainnet_e58758 - phase0 processRandaoMixesReset	27.409 us/op	16.998 us/op	1.61
mainnet_e58758 - phase0 processHistoricalRootsUpdate	4.7810 us/op	2.9770 us/op	1.61
mainnet_e58758 - phase0 processParticipationRecordUpdates	23.816 us/op	10.288 us/op	2.31
mainnet_e58758 - phase0 afterProcessEpoch	172.26 ms/op	174.58 ms/op	0.99
phase0 processEffectiveBalanceUpdates - 250000 normalcase	2.4514 ms/op	3.2537 ms/op	0.75
phase0 processEffectiveBalanceUpdates - 250000 worstcase 0.5	2.5295 ms/op	2.5699 ms/op	0.98
altair processInactivityUpdates - 250000 normalcase	52.870 ms/op	41.727 ms/op	1.27
altair processInactivityUpdates - 250000 worstcase	43.715 ms/op	54.246 ms/op	0.81
phase0 processRegistryUpdates - 250000 normalcase	31.241 us/op	16.894 us/op	1.85
phase0 processRegistryUpdates - 250000 badcase_full_deposits	486.77 us/op	502.21 us/op	0.97
phase0 processRegistryUpdates - 250000 worstcase 0.5	241.02 ms/op	239.85 ms/op	1.00
altair processRewardsAndPenalties - 250000 normalcase	141.00 ms/op	154.72 ms/op	0.91
altair processRewardsAndPenalties - 250000 worstcase	101.13 ms/op	101.19 ms/op	1.00
phase0 getAttestationDeltas - 250000 normalcase	13.299 ms/op	14.529 ms/op	0.92
phase0 getAttestationDeltas - 250000 worstcase	14.349 ms/op	15.659 ms/op	0.92
phase0 processSlashings - 250000 worstcase	6.2555 ms/op	6.0293 ms/op	1.04
altair processSyncCommitteeUpdates - 250000	315.62 ms/op	310.30 ms/op	1.02
BeaconState.hashTreeRoot - No change	564.00 ns/op	534.00 ns/op	1.06
BeaconState.hashTreeRoot - 1 full validator	75.337 us/op	61.536 us/op	1.22
BeaconState.hashTreeRoot - 32 full validator	702.02 us/op	647.60 us/op	1.08
BeaconState.hashTreeRoot - 512 full validator	7.2041 ms/op	6.5905 ms/op	1.09
BeaconState.hashTreeRoot - 1 validator.effectiveBalance	88.353 us/op	77.409 us/op	1.14
BeaconState.hashTreeRoot - 32 validator.effectiveBalance	1.2164 ms/op	1.0830 ms/op	1.12
BeaconState.hashTreeRoot - 512 validator.effectiveBalance	16.350 ms/op	14.285 ms/op	1.14
BeaconState.hashTreeRoot - 1 balances	71.784 us/op	60.958 us/op	1.18
BeaconState.hashTreeRoot - 32 balances	641.21 us/op	566.91 us/op	1.13
BeaconState.hashTreeRoot - 512 balances	5.9001 ms/op	5.1549 ms/op	1.14
BeaconState.hashTreeRoot - 250000 balances	105.56 ms/op	104.22 ms/op	1.01
aggregationBits - 2048 els - zipIndexesInBitList	31.818 us/op	28.894 us/op	1.10
regular array get 100000 times	55.328 us/op	74.653 us/op	0.74
wrappedArray get 100000 times	56.284 us/op	78.414 us/op	0.72
arrayWithProxy get 100000 times	35.359 ms/op	32.894 ms/op	1.07
ssz.Root.equals	594.00 ns/op	487.00 ns/op	1.22
byteArrayEquals	564.00 ns/op	493.00 ns/op	1.14
shuffle list - 16384 els	11.873 ms/op	12.700 ms/op	0.93
shuffle list - 250000 els	173.98 ms/op	184.69 ms/op	0.94
processSlot - 1 slots	16.530 us/op	11.663 us/op	1.42
processSlot - 32 slots	2.1382 ms/op	1.7728 ms/op	1.21
getEffectiveBalanceIncrementsZeroInactive - 250000 vs - 7PWei	415.58 us/op	407.34 us/op	1.02
getCommitteeAssignments - req 1 vs - 250000 vc	5.3033 ms/op	5.4755 ms/op	0.97
getCommitteeAssignments - req 100 vs - 250000 vc	7.3257 ms/op	7.2944 ms/op	1.00
getCommitteeAssignments - req 1000 vs - 250000 vc	7.8133 ms/op	7.8240 ms/op	1.00
computeProposers - vc 250000	20.515 ms/op	20.457 ms/op	1.00
computeEpochShuffling - vc 250000	169.11 ms/op	182.19 ms/op	0.93
getNextSyncCommittee - vc 250000	298.05 ms/op	297.07 ms/op	1.00

by benchmarkbot/action

[dapplion]

For some reason build:docs script breaks in this PR https://github.com/ChainSafe/lodestar/runs/7339174005?check_suite_focus=true

@chainsafe/lodestar: Error [ERR_LOADER_CHAIN_INCOMPLETE]: "file:///home/runner/work/lodestar/lodestar/node_modules/ts-node/child-loader.mjs 'resolve'" did not call the next hook in its chain and did not explicitly signal a short circuit. If this is intentional, include `shortCircuit: true` in the hook's return.
@chainsafe/lodestar:     at new NodeError (node:internal/errors:387:5)
@chainsafe/lodestar:     at ESMLoader.resolve (node:internal/modules/esm/loader:860:13)
@chainsafe/lodestar:     at async ESMLoader.getModuleJob (node:internal/modules/esm/loader:439:7)
@chainsafe/lodestar:     at async Promise.all (index 0)
@chainsafe/lodestar:     at async ESMLoader.import (node:internal/modules/esm/loader:541:24)
@chainsafe/lodestar:     at async loadESM (node:internal/process/esm_loader:83:5)
@chainsafe/lodestar:     at async handleMainPromise (node:internal/modules/run_main:65:12) {
@chainsafe/lodestar:   code: 'ERR_LOADER_CHAIN_INCOMPLETE'
@chainsafe/lodestar: }

Running locally:

• Node 16: it works
• Node 18: TBD

Fix:

• Blocked by Fix build:refdocs script in cli package #4297

[wemeetagain]

why is this configurable? because it was only used during debugging? and in normal practice it just adds overhead?

[wemeetagain]

Seems like it could/should just be 'always on'

[tuyennhv]

@wemeetagain yeah I think we can turn it on when:

• code is unchanged for a while
• sync through prater
• sync through mainnet
• other testnets?