test moc inlining

[chenyan-dfinity]

base: old metering, without wasm-opt

[github-actions[bot]]

Note Diffing the performance result against the published result from main branch. Unchanged benchmarks are omitted.

Map

	binary_size	generate 1m	max mem	batch_get 50	batch_put 50	batch_remove 50
hashmap	167_316 ($\textcolor{red}{5.10\%}$)	9_230_374_182 ($\textcolor{red}{21.63\%}$)	61_987_732	382_826 ($\textcolor{red}{20.03\%}$)	7_246_862_420 ($\textcolor{red}{20.62\%}$)	411_340 ($\textcolor{red}{20.49\%}$)
triemap	171_390 ($\textcolor{red}{5.75\%}$)	16_198_090_366 ($\textcolor{red}{28.16\%}$)	74_216_052	309_052 ($\textcolor{red}{23.42\%}$)	782_396 ($\textcolor{red}{28.03\%}$)	769_026 ($\textcolor{red}{27.97\%}$)
rbtree	171_502 ($\textcolor{red}{5.60\%}$)	7_930_546_839 ($\textcolor{red}{28.72\%}$)	57_995_940	137_477 ($\textcolor{red}{18.31\%}$)	357_040 ($\textcolor{red}{28.78\%}$)	401_036 ($\textcolor{red}{32.72\%}$)
splay	166_269 ($\textcolor{red}{5.43\%}$)	15_696_985_410 ($\textcolor{red}{27.77\%}$)	53_995_876	752_855 ($\textcolor{red}{27.79\%}$)	790_594 ($\textcolor{red}{27.59\%}$)	1_106_735 ($\textcolor{red}{27.12\%}$)
btree	243_933 ($\textcolor{red}{13.73\%}$)	11_541_962_095 ($\textcolor{red}{26.28\%}$)	31_103_892	400_549 ($\textcolor{red}{27.48\%}$)	543_692 ($\textcolor{red}{26.54\%}$)	612_045 ($\textcolor{red}{27.29\%}$)
zhenya_hashmap	184_431 ($\textcolor{red}{9.49\%}$)	3_538_128_781 ($\textcolor{red}{25.37\%}$)	65_987_480	93_396 ($\textcolor{red}{27.84\%}$)	112_641 ($\textcolor{red}{25.56\%}$)	129_949 ($\textcolor{red}{19.80\%}$)
btreemap_rs	446_267 ($\textcolor{red}{0.02\%}$)	1_797_752_179 ($\textcolor{red}{9.79\%}$)	13_762_560	74_544 ($\textcolor{red}{13.42\%}$)	126_136 ($\textcolor{red}{12.52\%}$)	92_839 ($\textcolor{red}{13.30\%}$)
imrc_hashmap_rs	446_166 ($\textcolor{red}{0.01\%}$)	2_571_892_333 ($\textcolor{red}{7.66\%}$)	122_454_016	38_956 ($\textcolor{red}{17.34\%}$)	179_095 ($\textcolor{red}{9.64\%}$)	115_561 ($\textcolor{red}{18.60\%}$)
hashmap_rs	439_346 ($\textcolor{red}{0.01\%}$)	447_664_894 ($\textcolor{red}{8.11\%}$)	36_536_320	22_228 ($\textcolor{red}{29.80\%}$)	27_664 ($\textcolor{red}{24.45\%}$)	25_290 ($\textcolor{red}{24.44\%}$)

Priority queue

	binary_size	heapify 1m	max mem	pop_min 50	put 50
heap	158_536 ($\textcolor{red}{3.98\%}$)	6_553_514_370 ($\textcolor{red}{26.01\%}$)	29_995_836	720_429 ($\textcolor{red}{25.49\%}$)	265_478 ($\textcolor{red}{27.04\%}$)
heap_rs	437_278 ($\textcolor{red}{0.02\%}$)	142_914_793 ($\textcolor{red}{14.02\%}$)	9_109_504	59_850 ($\textcolor{red}{11.13\%}$)	23_726 ($\textcolor{red}{28.08\%}$)

Growable array

	binary_size	generate 5k	max mem	batch_get 500	batch_put 500	batch_remove 500
buffer	171_769 ($\textcolor{red}{6.32\%}$)	2_839_066 ($\textcolor{red}{23.62\%}$)	65_508	108_809 ($\textcolor{red}{29.15\%}$)	896_597 ($\textcolor{red}{21.08\%}$)	186_809 ($\textcolor{red}{31.33\%}$)
vector	172_652 ($\textcolor{red}{7.35\%}$)	2_430_538 ($\textcolor{red}{28.59\%}$)	24_764	170_991 ($\textcolor{red}{24.46\%}$)	232_826 ($\textcolor{red}{29.81\%}$)	231_638 ($\textcolor{red}{27.68\%}$)
vec_rs	435_834 ($\textcolor{red}{0.01\%}$)	290_143 ($\textcolor{red}{8.96\%}$)	655_360	17_605 ($\textcolor{red}{33.16\%}$)	31_014 ($\textcolor{red}{20.92\%}$)	25_400 ($\textcolor{red}{22.22\%}$)

Statistics

• binary_size: 4.49% [2.51%, 6.46%]
• max_mem: no change
• cycles: 22.91% [21.35%, 24.47%]

SHA-2

	binary_size	SHA-256	SHA-512	account_id	neuron_id
Motoko	230_070 ($\textcolor{red}{17.26\%}$)	326_935_518 ($\textcolor{red}{14.46\%}$)	298_093_814 ($\textcolor{red}{16.60\%}$)	41_503 ($\textcolor{red}{19.45\%}$)	29_780 ($\textcolor{red}{25.71\%}$)
Rust	528_234 ($\textcolor{red}{0.01\%}$)	82_789_387 ($\textcolor{red}{0.34\%}$)	56_794_263 ($\textcolor{red}{0.47\%}$)	50_651 ($\textcolor{red}{16.93\%}$)	53_532 ($\textcolor{red}{17.71\%}$)

Certified map

	binary_size	generate 10k	max mem	inc	witness
Motoko	232_943 ($\textcolor{red}{13.19\%}$)	5_792_324_783 ($\textcolor{red}{14.70\%}$)	3_429_924	687_589 ($\textcolor{red}{14.91\%}$)	445_626 ($\textcolor{red}{23.36\%}$)
Rust	469_955 ($\textcolor{red}{0.02\%}$)	6_359_442_714 ($\textcolor{red}{2.03\%}$)	1_081_344	1_012_174 ($\textcolor{red}{2.36\%}$)	305_119 ($\textcolor{red}{4.88\%}$)

Statistics

• binary_size: 7.62% [-2.90%, 18.14%]
• max_mem: no change
• cycles: 12.42% [8.31%, 16.54%]

Basic DAO

	binary_size	init	transfer_token	submit_proposal	vote_proposal
Motoko	301_252 ($\textcolor{red}{8.53\%}$)	50_473 ($\textcolor{red}{27.10\%}$)	24_519 ($\textcolor{red}{37.88\%}$)	20_380 ($\textcolor{red}{47.35\%}$)	21_890 ($\textcolor{red}{40.23\%}$)
Rust	763_017 ($\textcolor{red}{0.02\%}$)	552_075 ($\textcolor{red}{14.02\%}$)	105_203 ($\textcolor{red}{18.14\%}$)	128_753 ($\textcolor{red}{19.48\%}$)	139_539 ($\textcolor{red}{17.55\%}$)

DIP721 NFT

	binary_size	init	mint_token	transfer_token
Motoko	244_819 ($\textcolor{red}{6.47\%}$)	19_098 ($\textcolor{red}{44.50\%}$)	31_854 ($\textcolor{red}{33.96\%}$)	9_565 ($\textcolor{red}{81.60\%}$)
Rust	828_238 ($\textcolor{red}{0.02\%}$)	146_257 ($\textcolor{red}{13.96\%}$)	380_260 ($\textcolor{red}{14.39\%}$)	93_763 ($\textcolor{red}{18.36\%}$)

Statistics

• binary_size: 3.76% [-1.42%, 8.94%]
• max_mem: no change
• cycles: 30.61% [21.65%, 39.57%]

Heartbeat

	binary_size	heartbeat
Motoko	145_761 ($\textcolor{red}{2.57\%}$)	20_036 ($\textcolor{red}{157.63\%}$)
Rust	25_650 ($\textcolor{red}{0.02\%}$)	1_179 ($\textcolor{red}{151.92\%}$)

Timer

	binary_size	setTimer	cancelTimer
Motoko	154_792 ($\textcolor{red}{3.67\%}$)	54_057 ($\textcolor{red}{228.20\%}$)	4_942 ($\textcolor{red}{167.57\%}$)
Rust	470_693 ($\textcolor{red}{0.02\%}$)	69_727 ($\textcolor{red}{56.66\%}$)	11_405 ($\textcolor{red}{47.87\%}$)

Statistics

• binary_size: 1.85% [-9.68%, 13.38%]
• max_mem: no change
• cycles: 125.07% [21.86%, 228.28%]

Garbage Collection

	generate 800k	max mem	batch_get 50	batch_put 50	batch_remove 50
default	1_338_231_405 ($\textcolor{red}{32.20\%}$)	59_396_776	118 ($\textcolor{red}{136.00\%}$)	118 ($\textcolor{red}{136.00\%}$)	118 ($\textcolor{red}{136.00\%}$)
copying	1_338_231_287 ($\textcolor{red}{32.20\%}$)	59_396_776	1_337_913_569 ($\textcolor{red}{32.17\%}$)	1_338_002_371 ($\textcolor{red}{32.17\%}$)	1_337_919_144 ($\textcolor{red}{32.17\%}$)
compacting	1_911_420_608 ($\textcolor{red}{14.11\%}$)	59_396_776	1_473_824_186 ($\textcolor{red}{13.99\%}$)	1_756_485_066 ($\textcolor{red}{14.63\%}$)	1_787_369_954 ($\textcolor{red}{14.69\%}$)
generational	2_891_818_643 ($\textcolor{red}{15.76\%}$)	59_405_240	1_141_865_993 ($\textcolor{red}{16.81\%}$)	1_217_376 ($\textcolor{red}{16.50\%}$)	1_117_840 ($\textcolor{red}{16.39\%}$)
incremental	33_436_719 ($\textcolor{red}{3.45\%}$)	1_136_155_048	333_734_166 ($\textcolor{red}{14.98\%}$)	336_829_512 ($\textcolor{red}{14.98\%}$)	336_860_690 ($\textcolor{red}{14.98\%}$)

Actor class

	binary size	put new bucket	put existing bucket	get
Map	311_868 ($\textcolor{red}{4.76\%}$)	818_044 ($\textcolor{red}{13.63\%}$)	16_950 ($\textcolor{red}{249.20\%}$)	17_406 ($\textcolor{red}{229.16\%}$)

Statistics

• binary_size: no change
• max_mem: no change
• cycles: 51.54% [26.87%, 76.20%]

Publisher & Subscriber

	pub_binary_size	sub_binary_size	subscribe_caller	subscribe_callee	publish_caller	publish_callee
Motoko	172_768 ($\textcolor{red}{3.59\%}$)	155_860 ($\textcolor{red}{2.64\%}$)	29_874 ($\textcolor{red}{95.04\%}$)	12_492 ($\textcolor{red}{42.64\%}$)	23_937 ($\textcolor{red}{116.78\%}$)	6_816 ($\textcolor{red}{76.03\%}$)
Rust	511_870 ($\textcolor{red}{0.02\%}$)	565_407 ($\textcolor{red}{0.01\%}$)	71_728 ($\textcolor{red}{35.43\%}$)	44_318 ($\textcolor{red}{24.84\%}$)	95_767 ($\textcolor{red}{25.08\%}$)	53_941 ($\textcolor{red}{18.82\%}$)

Statistics

• binary_size: 1.56% [-0.59%, 3.72%]

• max_mem: no change

• cycles: 54.33% [29.65%, 79.02%]

  Overall Statistics

• binary_size: 4.22% [2.67%, 5.78%]

• max_mem: no change

• cycles: 33.99% [27.49%, 40.50%]

[github-actions[bot]]

Note The flamegraph link only works after you merge. Unchanged benchmarks are omitted.

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.

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.

💎 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.

Note

• The Candid interface of the benchmark is minimal, therefore the serialization cost is negligible in this measurement.
• Due to the instrumentation overhead and cycle limit, we cannot profile computations with large collections. Hopefully, when deterministic time slicing is ready, we can measure the performance on larger memory footprint.
• hashmap uses amortized data structure. When the initial capacity is reached, it has to copy the whole array, thus the cost of batch_put 50 is much higher than other data structures.
• btree comes from mops.one/stableheapbtreemap.
• zhenya_hashmap comes from mops.one/map.
• vector comes from mops.one/vector. Compare with buffer, put has better worst case time and space complexity ($O(\sqrt{n})$ vs $O(n)$); get has a slightly larger constant overhead.
• hashmap_rs uses the fxhash crate, which is the same as std::collections::HashMap, but with a deterministic hasher. This ensures reproducible result.
• imrc_hashmap_rs uses the im-rc crate, which is the immutable version hashmap in Rust.

Map

	binary_size	generate 1m	max mem	batch_get 50	batch_put 50	batch_remove 50
hashmap	167_316	9_230_374_182	61_987_732	382_826	7_246_862_420	411_340
triemap	171_390	16_198_090_366	74_216_052	309_052	782_396	769_026
rbtree	171_502	7_930_546_839	57_995_940	137_477	357_040	401_036
splay	166_269	15_696_985_410	53_995_876	752_855	790_594	1_106_735
btree	243_933	11_541_962_095	31_103_892	400_549	543_692	612_045
zhenya_hashmap	184_431	3_538_128_781	65_987_480	93_396	112_641	129_949
btreemap_rs	446_267	1_797_752_179	13_762_560	74_544	126_136	92_839
imrc_hashmap_rs	446_166	2_571_892_333	122_454_016	38_956	179_095	115_561
hashmap_rs	439_346	447_664_894	36_536_320	22_228	27_664	25_290

Priority queue

	binary_size	heapify 1m	max mem	pop_min 50	put 50
heap	158_536	6_553_514_370	29_995_836	720_429	265_478	686_570
heap_rs	437_278	142_914_793	9_109_504	59_850	23_726	60_072

Growable array

	binary_size	generate 5k	max mem	batch_get 500	batch_put 500	batch_remove 500
buffer	171_769	2_839_066	65_508	108_809	896_597	186_809
vector	172_652	2_430_538	24_764	170_991	232_826	231_638
vec_rs	435_834	290_143	655_360	17_605	31_014	25_400

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.

SHA-2

	binary_size	SHA-256	SHA-512	account_id	neuron_id
Motoko	230_070	326_935_518	298_093_814	41_503	29_780
Rust	528_234	82_789_387	56_794_263	50_651	53_532

Certified map

	binary_size	generate 10k	max mem	inc	witness
Motoko	232_943	5_792_324_783	3_429_924	687_589	445_626
Rust	469_955	6_359_442_714	1_081_344	1_012_174	305_119

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

Note

• The cost difference is mainly due to the Candid serialization cost.
• Motoko statically compiles/specializes the serialization code for each method, whereas in Rust, we use serde to dynamically deserialize data based on data on the wire.
• We could improve the performance on the Rust side by using parser combinators. But it is a challenge to maintain the ergonomics provided by serde.
• For real-world applications, we tend to send small data for each endpoint, which makes the Candid overhead in Rust tolerable.

Basic DAO

	binary_size	init	transfer_token	submit_proposal	vote_proposal
Motoko	301_252	50_473	24_519	20_380	21_890
Rust	763_017	552_075	105_203	128_753	139_539

DIP721 NFT

	binary_size	init	mint_token	transfer_token
Motoko	244_819	19_098	31_854	9_565
Rust	828_238	146_257	380_260	93_763

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	145_761	20_036
Rust	25_650	1_179

Timer

	binary_size	setTimer	cancelTimer
Motoko	154_792	54_057	4_942
Rust	470_693	69_727	11_405

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 800k	max mem	batch_get 50	batch_put 50	batch_remove 50
default	1_338_231_405	59_396_776	118	118	118
copying	1_338_231_287	59_396_776	1_337_913_569	1_338_002_371	1_337_919_144
compacting	1_911_420_608	59_396_776	1_473_824_186	1_756_485_066	1_787_369_954
generational	2_891_818_643	59_405_240	1_141_865_993	1_217_376	1_117_840
incremental	33_436_719	1_136_155_048	333_734_166	336_829_512	336_860_690

Actor class

	binary size	put new bucket	put existing bucket	get
Map	311_868	818_044	16_950	17_406

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	172_768	155_860	29_874	12_492	23_937	6_816
Rust	511_870	565_407	71_728	44_318	95_767	53_941