experiment: only sample compact words and use better 64-bit hash in zhenya test #108

crusso · 2024-02-16T10:47:43Z

No description provided.

collections/motoko/src/ZhenyaHashmap.mo

github-actions · 2024-02-16T13:01:27Z

Map

	binary_size	generate 1m	max mem	batch_get 50	batch_put 50	batch_remove 50	upgrade
hashmap	190_578 ($\textcolor{red}{0.51\%}$)	8_677_435_631 ($\textcolor{red}{3.66\%}$)	59_550_664 ($\textcolor{green}{-3.93\%}$)	334_166 ($\textcolor{green}{-3.12\%}$)	372_166 ($\textcolor{green}{-99.99\%}$)	361_464 ($\textcolor{green}{-2.63\%}$)	11_094_880_659 ($\textcolor{red}{0.62\%}$)
triemap	196_095 ($\textcolor{red}{0.30\%}$)	13_905_101_144 ($\textcolor{red}{0.36\%}$)	71_717_452 ($\textcolor{green}{-3.37\%}$)	249_369 ($\textcolor{green}{-2.05\%}$)	642_587 ($\textcolor{green}{-2.85\%}$)	645_747 ($\textcolor{green}{-0.78\%}$)	15_695_504_052 ($\textcolor{green}{-0.77\%}$)
rbtree	186_336 ($\textcolor{green}{-0.12\%}$)	7_037_815_635 ($\textcolor{green}{-1.26\%}$)	51_814_928 ($\textcolor{green}{-10.66\%}$)	118_641 ($\textcolor{red}{3.80\%}$)	318_954 ($\textcolor{red}{0.19\%}$)	332_694 ($\textcolor{red}{1.35\%}$)	6_872_277_147 ($\textcolor{green}{-4.14\%}$)
splay	191_049 ($\textcolor{red}{0.28\%}$)	13_185_262_200 ($\textcolor{green}{-0.47\%}$)	47_829_136 ($\textcolor{green}{-11.42\%}$)	621_734 ($\textcolor{green}{-1.10\%}$)	677_921 ($\textcolor{red}{2.47\%}$)	937_345 ($\textcolor{red}{1.67\%}$)	4_333_406_943 ($\textcolor{green}{-5.13\%}$)
btree	230_967 ($\textcolor{red}{0.48\%}$)	10_843_807_996 ($\textcolor{red}{5.63\%}$)	25_021_016 ($\textcolor{green}{-19.56\%}$)	379_413 ($\textcolor{red}{7.29\%}$)	492_873 ($\textcolor{red}{2.23\%}$)	572_201 ($\textcolor{red}{7.17\%}$)	2_856_105_607 ($\textcolor{green}{-8.87\%}$)
zhenya_hashmap	188_742 ($\textcolor{green}{-0.29\%}$)	1_710_509_220 ($\textcolor{green}{-33.46\%}$)	16_777_504 ($\textcolor{green}{-26.33\%}$)	50_761 ($\textcolor{green}{-15.67\%}$)	73_525 ($\textcolor{red}{4.83\%}$)	112_957 ($\textcolor{red}{37.00\%}$)	5_610_596_903 ($\textcolor{red}{69.73\%}$)
btreemap_rs	537_393	1_793_333_047	27_590_656	75_328	125_166	86_260	2_937_041_107
imrc_hashmap_rs	542_882	2_584_501_850	244_973_568	37_762	178_926	115_385	5_796_587_958
hashmap_rs	529_458	439_248_112	73_138_176	21_501	26_711	25_024	1_298_646_667

Priority queue

	binary_size	heapify 1m	max mem	pop_min 50	put 50	pop_min 50.1	upgrade
heap	167_286 ($\textcolor{green}{-0.13\%}$)	5_702_353_152 ($\textcolor{red}{0.08\%}$)	24_000_360 ($\textcolor{green}{-19.99\%}$)	635_926 ($\textcolor{red}{2.35\%}$)	234_265 ($\textcolor{red}{2.44\%}$)	606_194 ($\textcolor{red}{2.36\%}$)	3_193_825_390 ($\textcolor{green}{-3.50\%}$)
heap_rs	525_853	139_669_830	18_284_544	57_419	23_051	57_545	510_960_192

Growable array

	binary_size	generate 5k	max mem	batch_get 500	batch_put 500	batch_remove 500	upgrade
buffer	174_288 ($\textcolor{red}{0.22\%}$)	2_799_267 ($\textcolor{red}{8.88\%}$)	65_644	100_512 ($\textcolor{red}{5.26\%}$)	866_454 ($\textcolor{red}{8.25\%}$)	187_512 ($\textcolor{red}{9.98\%}$)	3_141_338 ($\textcolor{red}{2.61\%}$)
vector	172_193 ($\textcolor{green}{-0.03\%}$)	2_048_943 ($\textcolor{red}{6.66\%}$)	24_580	133_138 ($\textcolor{red}{5.57\%}$)	195_575 ($\textcolor{red}{6.65\%}$)	184_331 ($\textcolor{red}{4.74\%}$)	4_685_196 ($\textcolor{green}{-0.22\%}$)
vec_rs	520_881	289_040	1_376_256	17_251	30_571	23_331	3_161_017

Stable structures

Statistics

binary_size: 0.14% [-0.04%, 0.31%]
max_mem: -13.61% [-19.97%, -7.25%]
cycles: 0.62% [-4.50%, 5.74%]

SHA-2

	binary_size	SHA-256	SHA-512	account_id	neuron_id
Motoko	195_513 ($\textcolor{green}{-0.50\%}$)	281_134_819 ($\textcolor{red}{2.97\%}$)	261_270_435 ($\textcolor{red}{0.55\%}$)	35_174 ($\textcolor{red}{2.33\%}$)	25_432 ($\textcolor{red}{2.15\%}$)
Rust	537_397	82_787_911	56_792_991	47_914	50_388

Certified map

	binary_size	generate 10k	max mem	inc	witness	upgrade
Motoko	245_076 ($\textcolor{green}{-0.04\%}$)	4_883_900_392 ($\textcolor{red}{2.53\%}$)	3_430_044	579_659 ($\textcolor{red}{2.57\%}$)	432_230 ($\textcolor{red}{7.50\%}$)	274_734_742 ($\textcolor{red}{0.17\%}$)
Rust	565_792	6_409_147_805	2_228_224	1_019_959	303_897	6_019_483_730

Statistics

binary_size: -0.27% [-1.71%, 1.16%]
max_mem: no change
cycles: 2.60% [1.11%, 4.08%]

Basic DAO

	binary_size	init	transfer_token	submit_proposal	vote_proposal	upgrade
Motoko	274_439 ($\textcolor{green}{-0.37\%}$)	510_973 ($\textcolor{red}{0.06\%}$)	22_488 ($\textcolor{red}{1.02\%}$)	18_681 ($\textcolor{red}{0.37\%}$)	19_761 ($\textcolor{red}{0.64\%}$)	157_965 ($\textcolor{red}{0.23\%}$)
Rust	849_921	599_916 ($\textcolor{green}{-0.01\%}$)	99_156	123_702	136_655	1_799_828 ($\textcolor{green}{-0.00\%}$)

DIP721 NFT

	binary_size	init	mint_token	transfer_token	upgrade
Motoko	220_659 ($\textcolor{green}{-0.88\%}$)	481_160 ($\textcolor{red}{0.00\%}$)	30_218 ($\textcolor{red}{1.37\%}$)	8_776	89_843 ($\textcolor{red}{0.43\%}$)
Rust	869_104	236_542	368_044	91_941	1_999_207 ($\textcolor{green}{-0.00\%}$)

Statistics

binary_size: -0.62% [-2.22%, 0.97%]
max_mem: no change
cycles: 0.37% [0.12%, 0.63%]

Heartbeat

	binary_size	heartbeat
Motoko	137_356 ($\textcolor{green}{-0.39\%}$)	19_515 ($\textcolor{red}{0.02\%}$)
Rust	23_637	480 ($\textcolor{green}{-56.83\%}$)

Timer

	binary_size	setTimer	cancelTimer
Motoko	145_843 ($\textcolor{green}{-0.31\%}$)	53_003 ($\textcolor{red}{2.61\%}$)	4_628 ($\textcolor{red}{0.39\%}$)
Rust	487_585	68_173	11_184

Statistics

binary_size: -0.31%
max_mem: no change
cycles: 1.50% [-5.51%, 8.51%]

Garbage Collection

	generate 700k	max mem	batch_get 50	batch_put 50	batch_remove 50
default	1_146_951_684 ($\textcolor{green}{-2.09\%}$)	50_288_680 ($\textcolor{green}{-3.27\%}$)	119	119	119
copying	1_146_951_566 ($\textcolor{green}{-2.09\%}$)	50_288_680 ($\textcolor{green}{-3.27\%}$)	1_146_669_222 ($\textcolor{green}{-2.09\%}$)	1_146_757_219 ($\textcolor{green}{-2.09\%}$)	1_146_671_127 ($\textcolor{green}{-2.09\%}$)
compacting	1_643_059_256 ($\textcolor{green}{-1.74\%}$)	50_288_680 ($\textcolor{green}{-3.27\%}$)	1_268_924_251 ($\textcolor{green}{-1.64\%}$)	1_509_402_797 ($\textcolor{green}{-1.57\%}$)	1_537_068_318 ($\textcolor{green}{-1.75\%}$)
generational	2_460_323_534 ($\textcolor{green}{-2.72\%}$)	50_297_144 ($\textcolor{green}{-3.27\%}$)	952_674_141 ($\textcolor{green}{-4.69\%}$)	1_247_801 ($\textcolor{red}{1.22\%}$)	1_112_579 ($\textcolor{red}{0.79\%}$)
incremental	29_503_170	979_645_844 ($\textcolor{green}{-0.63\%}$)	478_912_995 ($\textcolor{red}{0.20\%}$)	494_653_376 ($\textcolor{red}{0.21\%}$)	1_149_877_816 ($\textcolor{red}{5.27\%}$)

Actor class

	binary size	put new bucket	put existing bucket	get
Map	299_783 ($\textcolor{green}{-0.13\%}$)	815_147 ($\textcolor{green}{-0.11\%}$)	16_125 ($\textcolor{red}{0.16\%}$)	16_670 ($\textcolor{red}{0.16\%}$)

Statistics

binary_size: no change
max_mem: -2.75% [-3.87%, -1.62%]
cycles: -0.84% [-1.62%, -0.05%]

Publisher & Subscriber

	pub_binary_size	sub_binary_size	subscribe_caller	subscribe_callee	publish_caller	publish_callee
Motoko	161_484 ($\textcolor{green}{-0.28\%}$)	145_919 ($\textcolor{green}{-0.21\%}$)	28_597 ($\textcolor{red}{0.01\%}$)	11_963	22_870 ($\textcolor{red}{0.03\%}$)	6_454 ($\textcolor{red}{0.37\%}$)
Rust	519_866	570_028	68_903	42_634	92_131	51_818

Statistics

binary_size: -0.24% [-0.46%, -0.03%]
max_mem: no change
cycles: 0.14% [-0.21%, 0.48%]

Overall Statistics

binary_size: -0.09% [-0.25%, 0.08%]
max_mem: -9.08% [-13.50%, -4.66%]
cycles: 0.44% [-2.12%, 3.00%]

github-actions · 2024-02-16T13:01:28Z

Collection libraries

Measure different collection libraries written in both Motoko and Rust.
The library names with _rs suffix are written in Rust; the rest are written in Motoko.
The _stable and _stable_rs suffix represents that the library directly writes the state to stable memory using Region in Motoko and ic-stable-stuctures in Rust.

We use the same random number generator with fixed seed to ensure that all collections contain
the same elements, and the queries are exactly the same. Below we explain the measurements of each column in the table:

generate 1m. Insert 1m Nat64 integers into the collection. For Motoko collections, it usually triggers the GC; the rest of the column are not likely to trigger GC.
max mem. For Motoko, it reports rts_max_heap_size after generate call; For Rust, it reports the Wasm's memory page * 32Kb.
batch_get 50. Find 50 elements from the collection.
batch_put 50. Insert 50 elements to the collection.
batch_remove 50. Remove 50 elements from the collection.
upgrade. Upgrade the canister with the same Wasm module. For non-stable benchmarks, the map state is persisted by serializing and deserializing states into stable memory. For stable benchmarks, the upgrade takes no cycles, as the state is already in the stable memory.

💎 Takeaways

The platform only charges for instruction count. Data structures which make use of caching and locality have no impact on the cost.
We have a limit on the maximal cycles per round. This means asymptotic behavior doesn't matter much. We care more about the performance up to a fixed N. In the extreme cases, you may see an $O(10000 n\log n)$ algorithm hitting the limit, while an $O(n^2)$ algorithm runs just fine.
Amortized algorithms/GC may need to be more eager to avoid hitting the cycle limit on a particular round.
Rust costs more cycles to process complicated Candid data, but it is more efficient in performing core computations.

Map

	binary_size	generate 1m	max mem	batch_get 50	batch_put 50	batch_remove 50	upgrade
hashmap	190_578	8_677_435_631	59_550_664	334_166	372_166	361_464	11_094_880_659
triemap	196_095	13_905_101_144	71_717_452	249_369	642_587	645_747	15_695_504_052
rbtree	186_336	7_037_815_635	51_814_928	118_641	318_954	332_694	6_872_277_147
splay	191_049	13_185_262_200	47_829_136	621_734	677_921	937_345	4_333_406_943
btree	230_967	10_843_807_996	25_021_016	379_413	492_873	572_201	2_856_105_607
zhenya_hashmap	188_742	1_710_509_220	16_777_504	50_761	73_525	112_957	5_610_596_903
btreemap_rs	537_393	1_793_333_047	27_590_656	75_328	125_166	86_260	2_937_041_107
imrc_hashmap_rs	542_882	2_584_501_850	244_973_568	37_762	178_926	115_385	5_796_587_958
hashmap_rs	529_458	439_248_112	73_138_176	21_501	26_711	25_024	1_298_646_667

Priority queue

	binary_size	heapify 1m	max mem	pop_min 50	put 50	pop_min 50	upgrade
heap	167_286	5_702_353_152	24_000_360	635_926	234_265	606_194	3_193_825_390
heap_rs	525_853	139_669_830	18_284_544	57_419	23_051	57_545	510_960_192

Growable array

	binary_size	generate 5k	max mem	batch_get 500	batch_put 500	batch_remove 500	upgrade
buffer	174_288	2_799_267	65_644	100_512	866_454	187_512	3_141_338
vector	172_193	2_048_943	24_580	133_138	195_575	184_331	4_685_196
vec_rs	520_881	289_040	1_376_256	17_251	30_571	23_331	3_161_017

Stable structures

	binary_size	generate 50k	max mem	batch_get 50	batch_put 50	batch_remove 50	upgrade
btreemap_rs	537_393	76_200_333	2_555_904	64_886	97_044	85_272	126_265_270
btreemap_stable_rs	543_609	4_561_985_735	2_031_616	2_707_064	5_026_642	8_594_683	729_311
heap_rs	525_853	7_051_730	2_293_760	49_928	23_299	49_894	26_768_703
heap_stable_rs	506_559	271_553_517	458_752	2_294_851	238_596	2_277_771	729_317
vec_rs	520_881	3_079_382	2_293_760	17_251	18_421	17_719	24_671_551
vec_stable_rs	503_829	63_394_912	458_752	62_491	79_685	81_633	729_320

Cryptographic libraries

Measure different cryptographic libraries written in both Motoko and Rust.

SHA-2 benchmarks
- SHA-256/SHA-512. Compute the hash of a 1M Wasm binary.
- account_id. Compute the ledger account id from principal, based on SHA-224.
- neuron_id. Compute the NNS neuron id from principal, based on SHA-256.
Certified map. Merkle Tree for storing key-value pairs and generate witness according to the IC Interface Specification.
- generate 10k. Insert 10k 7-character word as both key and value into the certified map.
- max mem. For Motoko, it reports rts_max_heap_size after generate call; For Rust, it reports the Wasm's memory page * 32Kb.
- inc. Increment a counter and insert the counter value into the map.
- witness. Generate the root hash and a witness for the counter.
- upgrade. Upgrade the canister with the same Wasm. In Motoko, we use stable variable. In Rust, we convert the tree to a vector before serialization.

SHA-2

	binary_size	SHA-256	SHA-512	account_id	neuron_id
Motoko	195_513	281_134_819	261_270_435	35_174	25_432
Rust	537_397	82_787_911	56_792_991	47_914	50_388

Certified map

	binary_size	generate 10k	max mem	inc	witness	upgrade
Motoko	245_076	4_883_900_392	3_430_044	579_659	432_230	274_734_742
Rust	565_792	6_409_147_805	2_228_224	1_019_959	303_897	6_019_483_730

Sample Dapps

Measure the performance of some typical dapps:

Basic DAO,
with heartbeat disabled to make profiling easier. We have a separate benchmark to measure heartbeat performance.
DIP721 NFT

Basic DAO

	binary_size	init	transfer_token	submit_proposal	vote_proposal	upgrade
Motoko	274_439	510_973	22_488	18_681	19_761	157_965
Rust	849_921	599_916	99_156	123_702	136_655	1_799_828

DIP721 NFT

	binary_size	init	mint_token	transfer_token	upgrade
Motoko	220_659	481_160	30_218	8_776	89_843
Rust	869_104	236_542	368_044	91_941	1_999_207

Heartbeat / Timer

Measure the cost of empty heartbeat and timer job.

setTimer measures both the setTimer(0) method and the execution of empty job.
It is not easy to reliably capture the above events in one flamegraph, as the implementation detail
of the replica can affect how we measure this. Typically, a correct flamegraph contains both setTimer and canister_global_timer function. If it's not there, we may need to adjust the script.

Heartbeat

	binary_size	heartbeat
Motoko	137_356	19_515
Rust	23_637	480

Timer

	binary_size	setTimer	cancelTimer
Motoko	145_843	53_003	4_628
Rust	487_585	68_173	11_184

Motoko Specific Benchmarks

Measure various features only available in Motoko.

Garbage Collection. Measure Motoko garbage collection cost using the Triemap benchmark. The max mem column reports rts_max_heap_size after generate call. The cycle cost numbers reported here are garbage collection cost only. Some flamegraphs are truncated due to the 2M log size limit. The dfx/ic-wasm optimizer is disabled for the garbage collection test cases due to how the optimizer affects function names, making profiling trickier.
- default. Compile with the default GC option. With the current GC scheduler, generate will trigger the copying GC. The rest of the methods will not trigger GC.
- copying. Compile with --force-gc --copying-gc.
- compacting. Compile with --force-gc --compacting-gc.
- generational. Compile with --force-gc --generational-gc.
- incremental. Compile with --force-gc --incremental-gc.
Actor class. Measure the cost of spawning actor class, using the Actor classes example.

Garbage Collection

	generate 700k	max mem	batch_get 50	batch_put 50	batch_remove 50
default	1_146_951_684	50_288_680	119	119	119
copying	1_146_951_566	50_288_680	1_146_669_222	1_146_757_219	1_146_671_127
compacting	1_643_059_256	50_288_680	1_268_924_251	1_509_402_797	1_537_068_318
generational	2_460_323_534	50_297_144	952_674_141	1_247_801	1_112_579
incremental	29_503_170	979_645_844	478_912_995	494_653_376	1_149_877_816

Actor class

	binary size	put new bucket	put existing bucket	get
Map	299_783	815_147	16_125	16_670

Publisher & Subscriber

Measure the cost of inter-canister calls from the Publisher & Subscriber example.

	pub_binary_size	sub_binary_size	subscribe_caller	subscribe_callee	publish_caller	publish_callee
Motoko	161_484	145_919	28_597	11_963	22_870	6_454
Rust	519_866	570_028	68_903	42_634	92_131	51_818

collections/motoko/src/hashmap.mo

collections/motoko/src/ZhenyaHashmap.mo

chenyan-dfinity and others added 13 commits May 2, 2023 17:26

perf from moc artifact

a38b430

fix

510bad2

Merge branch 'main' into moc-build

60451ee

trigger CI

48d0853

merge

756658e

merge with origin/main

bc4c7c7

Update .github/workflows/perf.yml

71cfcde

retrigger CI

81ed4c7

use better hash function

1b763c4

avoid boxing (changes baseline)

77a920e

restore RNG; return 28 high bits instead to avoid boxing and duplication

d54690b

insert spaces

bc41fe4

use better hash

9431cae

crusso commented Feb 16, 2024

View reviewed changes

collections/motoko/src/ZhenyaHashmap.mo Outdated Show resolved Hide resolved

Apply suggestions from code review

d15162e

crusso commented Feb 16, 2024

View reviewed changes

collections/motoko/src/hashmap.mo Outdated Show resolved Hide resolved

simplify hash to just use low 28 bits

f6b671d

crusso commented Feb 16, 2024

View reviewed changes

collections/motoko/src/ZhenyaHashmap.mo Outdated Show resolved Hide resolved

Apply suggestions from code review

ae72433

crusso commented Feb 16, 2024

View reviewed changes

collections/motoko/src/ZhenyaHashmap.mo Outdated Show resolved Hide resolved

crusso added 3 commits February 16, 2024 17:16

Update collections/motoko/src/ZhenyaHashmap.mo

e52f850

simplify hash to just use low 27 bits

7b87f91

simplify hash to just use low 27 bits

b0f8341

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

experiment: only sample compact words and use better 64-bit hash in zhenya test #108

experiment: only sample compact words and use better 64-bit hash in zhenya test #108

crusso commented Feb 16, 2024

github-actions bot commented Feb 16, 2024 •

edited

Loading

github-actions bot commented Feb 16, 2024 •

edited

Loading

Environment

Environment

Environment

Environment

Environment

Environment

experiment: only sample compact words and use better 64-bit hash in zhenya test #108

Are you sure you want to change the base?

experiment: only sample compact words and use better 64-bit hash in zhenya test #108

Conversation

crusso commented Feb 16, 2024

github-actions bot commented Feb 16, 2024 • edited Loading

Map

Priority queue

Growable array

Stable structures

Statistics

SHA-2

Certified map

Statistics

Basic DAO

DIP721 NFT

Statistics

Heartbeat

Timer

Statistics

Garbage Collection

Actor class

Statistics

Publisher & Subscriber

Statistics

Overall Statistics

github-actions bot commented Feb 16, 2024 • edited Loading

Collection libraries

💎 Takeaways

Map

Priority queue

Growable array

Stable structures

Environment

Cryptographic libraries

SHA-2

Certified map

Environment

Sample Dapps

Basic DAO

DIP721 NFT

Environment

Heartbeat / Timer

Heartbeat

Timer

Environment

Motoko Specific Benchmarks

Garbage Collection

Actor class

Environment

Publisher & Subscriber

Environment

github-actions bot commented Feb 16, 2024 •

edited

Loading

github-actions bot commented Feb 16, 2024 •

edited

Loading