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	upgrade
hashmap	189_929	8_184_618_025	56_000_256	342_784	6_462_528_122	368_420	10_728_193_099
triemap	195_472	13_661_315_924	68_228_576	252_649	657_794	648_084	15_499_470_884
rbtree	185_907	7_009_043_570	52_000_464	116_348	318_320	330_226	6_870_900_152
splay	190_457	13_157_617_583	48_000_400	631_329	662_998	928_144	4_308_925_798
btree	230_321	10_223_929_607	25_108_416	357_912	485_794	539_490	2_861_974_825
zhenya_hashmap	188_894	2_360_638_679	16_777_504	58_204	66_552	79_675	3_018_208_083
btreemap_rs	555_925 ($\textcolor{red}{0.41\%}$)	1_792_610_876 ($\textcolor{green}{-0.00\%}$)	27_590_656	73_295 ($\textcolor{green}{-2.81\%}$)	123_191 ($\textcolor{green}{-1.63\%}$)	85_043 ($\textcolor{green}{-1.98\%}$)	3_204_151_479 ($\textcolor{green}{-0.00\%}$)
imrc_hashmap_rs	556_123 ($\textcolor{red}{0.08\%}$)	2_584_010_729 ($\textcolor{green}{-0.02\%}$)	244_908_032 ($\textcolor{green}{-0.03\%}$)	35_545 ($\textcolor{green}{-6.51\%}$)	194_927 ($\textcolor{red}{8.79\%}$)	93_154 ($\textcolor{green}{-19.73\%}$)	6_272_274_871 ($\textcolor{red}{0.15\%}$)
hashmap_rs	544_570 ($\textcolor{red}{0.48\%}$)	439_245_801 ($\textcolor{green}{-0.90\%}$)	73_138_176	20_008 ($\textcolor{green}{-7.22\%}$)	24_930 ($\textcolor{green}{-6.45\%}$)	23_243 ($\textcolor{green}{-6.88\%}$)	1_565_611_027 ($\textcolor{green}{-0.00\%}$)

Priority queue

	binary_size	heapify 1m	max mem	pop_min 50	put 50	pop_min 50.1	upgrade
heap	166_903	5_554_617_018	24_000_360	621_690	227_224	592_588	3_189_831_485
heap_rs	540_668 ($\textcolor{red}{1.43\%}$)	139_667_629 ($\textcolor{green}{-0.00\%}$)	18_284_544	55_880 ($\textcolor{green}{-2.71\%}$)	21_271 ($\textcolor{green}{-7.48\%}$)	55_765 ($\textcolor{green}{-2.99\%}$)	648_921_342 ($\textcolor{green}{-0.00\%}$)

Growable array

	binary_size	generate 5k	max mem	batch_get 500	batch_put 500	batch_remove 500	upgrade
buffer	173_903	2_601_059	65_644	95_506	803_474	173_506	3_091_310
vector	171_932	1_952_689	24_580	126_130	186_485	176_123	4_675_192
vec_rs	538_516 ($\textcolor{red}{0.97\%}$)	286_776 ($\textcolor{green}{-0.91\%}$)	1_376_256	15_714 ($\textcolor{green}{-9.08\%}$)	28_793 ($\textcolor{green}{-5.67\%}$)	21_553 ($\textcolor{green}{-7.44\%}$)	3_805_750 ($\textcolor{green}{-0.55\%}$)

Stable structures

	binary_size	generate 50k	max mem	batch_get 50	batch_put 50	batch_remove 50	upgrade
btreemap_rs	555_925 ($\textcolor{red}{0.41\%}$)	76_161_521 ($\textcolor{green}{-0.00\%}$)	2_555_904	62_853 ($\textcolor{green}{-3.26\%}$)	95_118 ($\textcolor{green}{-2.10\%}$)	84_048 ($\textcolor{green}{-2.00\%}$)	139_592_121 ($\textcolor{green}{-0.02\%}$)
btreemap_stable_rs	558_089 ($\textcolor{red}{0.48\%}$)	4_564_070_850 ($\textcolor{green}{-0.00\%}$)	2_031_616	2_709_647 ($\textcolor{red}{0.13\%}$)	5_031_627 ($\textcolor{green}{-0.05\%}$)	8_581_114 ($\textcolor{red}{0.03\%}$)	729_344 ($\textcolor{red}{0.00\%}$)
heap_rs	540_668 ($\textcolor{red}{1.43\%}$)	7_049_481 ($\textcolor{green}{-0.04\%}$)	2_293_760	48_389 ($\textcolor{green}{-3.12\%}$)	21_519 ($\textcolor{green}{-7.40\%}$)	48_114 ($\textcolor{green}{-3.45\%}$)	33_629_765 ($\textcolor{green}{-0.09\%}$)
heap_stable_rs	521_160 ($\textcolor{red}{0.65\%}$)	277_865_060 ($\textcolor{red}{2.04\%}$)	458_752	2_405_453 ($\textcolor{red}{4.28\%}$)	242_829 ($\textcolor{red}{1.47\%}$)	2_387_316 ($\textcolor{red}{4.27\%}$)	729_349
vec_rs	538_516 ($\textcolor{red}{0.97\%}$)	3_077_152 ($\textcolor{green}{-0.09\%}$)	2_293_760	15_714 ($\textcolor{green}{-9.08\%}$)	16_643 ($\textcolor{green}{-9.43\%}$)	15_941 ($\textcolor{green}{-9.80\%}$)	31_301_345 ($\textcolor{green}{-0.07\%}$)
vec_stable_rs	521_344 ($\textcolor{red}{1.18\%}$)	63_342_333 ($\textcolor{green}{-0.00\%}$)	458_752	64_876 ($\textcolor{red}{0.69\%}$)	78_806 ($\textcolor{green}{-1.16\%}$)	84_168 ($\textcolor{red}{0.65\%}$)	729_349 ($\textcolor{green}{-0.00\%}$)

Statistics

binary_size: 0.77% [0.52%, 1.02%]
max_mem: -0.03%
cycles: -2.22% [-3.23%, -1.20%]

SHA-2

	binary_size	SHA-256	SHA-512	account_id	neuron_id
Motoko	193_686	267_743_355	247_834_501	33_636	24_532
Rust	541_887 ($\textcolor{red}{0.60\%}$)	82_781_801 ($\textcolor{green}{-0.01\%}$)	56_787_280 ($\textcolor{green}{-0.01\%}$)	41_044 ($\textcolor{green}{-14.41\%}$)	39_949 ($\textcolor{green}{-21.47\%}$)

Certified map

	binary_size	generate 10k	max mem	inc	witness	upgrade
Motoko	243_641	4_666_119_661	3_430_044	553_629	407_936	274_434_719
Rust	580_473 ($\textcolor{red}{0.17\%}$)	6_409_375_828 ($\textcolor{green}{-0.00\%}$)	2_228_224	1_020_376 ($\textcolor{green}{-0.02\%}$)	298_204 ($\textcolor{green}{-2.16\%}$)	6_026_009_495 ($\textcolor{red}{0.00\%}$)

Statistics

binary_size: 0.38% [-0.97%, 1.73%]
max_mem: no change
cycles: -4.76% [-10.38%, 0.86%]

Basic DAO

	binary_size	init	transfer_token	submit_proposal	vote_proposal	upgrade
Motoko	273_938	510_793 ($\textcolor{green}{-0.01\%}$)	22_382 ($\textcolor{red}{0.02\%}$)	18_539 ($\textcolor{green}{-0.31\%}$)	19_569 ($\textcolor{green}{-0.48\%}$)	157_946 ($\textcolor{green}{-0.01\%}$)
Rust	847_522 ($\textcolor{red}{0.06\%}$)	506_485 ($\textcolor{green}{-19.28\%}$)	88_938 ($\textcolor{green}{-11.75\%}$)	117_357 ($\textcolor{green}{-10.80\%}$)	112_347 ($\textcolor{green}{-19.37\%}$)	1_483_364 ($\textcolor{green}{-18.90\%}$)

DIP721 NFT

	binary_size	init	mint_token	transfer_token	upgrade
Motoko	220_403	481_158	30_204	8_764	89_833
Rust	879_920 ($\textcolor{red}{0.23\%}$)	203_509 ($\textcolor{green}{-16.30\%}$)	303_060 ($\textcolor{green}{-21.69\%}$)	71_233 ($\textcolor{green}{-21.57\%}$)	1_629_856 ($\textcolor{green}{-21.97\%}$)

Statistics

binary_size: 0.14% [-0.40%, 0.69%]
max_mem: no change
cycles: -11.60% [-16.07%, -7.13%]

Heartbeat

	binary_size	heartbeat
Motoko	137_183	19_507
Rust	23_657	480 ($\textcolor{green}{-56.83\%}$)

Timer

	binary_size	setTimer	cancelTimer
Motoko	145_619	51_778	4_626
Rust	502_754 ($\textcolor{red}{1.02\%}$)	63_379 ($\textcolor{green}{-7.26\%}$)	11_676 ($\textcolor{green}{-0.19\%}$)

Statistics

binary_size: 1.02%
max_mem: no change
cycles: -3.72% [-26.04%, 18.59%]

Publisher & Subscriber

	pub_binary_size	sub_binary_size	subscribe_caller	subscribe_callee	publish_caller	publish_callee
Motoko	161_290	145_741	28_593	11_963	22_854	6_446
Rust	535_946 ($\textcolor{red}{0.29\%}$)	573_818 ($\textcolor{red}{0.16\%}$)	57_488 ($\textcolor{green}{-17.70\%}$)	37_798 ($\textcolor{green}{-13.71\%}$)	71_349 ($\textcolor{green}{-24.24\%}$)	42_449 ($\textcolor{green}{-20.43\%}$)

Statistics

binary_size: 0.22% [-0.20%, 0.64%]
max_mem: no change
cycles: -19.02% [-24.25%, -13.79%]

Overall Statistics
binary_size: 0.61% [0.43%, 0.79%]
max_mem: -0.03%
cycles: -4.92% [-6.31%, -3.54%]

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. 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.

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 very large collections.

The upgrade column uses Candid for serializing stable data. In Rust, you may get better cycle cost by using a different serialization format. Another slowdown in Rust is that ic-stable-structures tends to be slower than the region memory in Motoko.

Different library has different ways for persisting data during upgrades, there are mainly three categories:

Use stable variable directly in Motoko: zhenya_hashmap, btree, vector

Expose and serialize external state (share/unshare in Motoko, candid::Encode in Rust): rbtree, heap, btreemap_rs, hashmap_rs, heap_rs, vector_rs

Use pre/post-upgrade hooks to convert data into an array: hashmap, splay, triemap, buffer, imrc_hashmap_rs

The stable benchmarks are much more expensive than their non-stable counterpart, because the stable memory API is much more expensive. The benefit is that they get fast upgrade. The upgrade still needs to parse the metadata when initializing the upgraded Wasm module.

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	upgrade
hashmap	189_929	8_184_618_025	56_000_256	342_784	6_462_528_122	368_420	10_728_193_099
triemap	195_472	13_661_315_924	68_228_576	252_649	657_794	648_084	15_499_470_884
rbtree	185_907	7_009_043_570	52_000_464	116_348	318_320	330_226	6_870_900_152
splay	190_457	13_157_617_583	48_000_400	631_329	662_998	928_144	4_308_925_798
btree	230_321	10_223_929_607	25_108_416	357_912	485_794	539_490	2_861_974_825
zhenya_hashmap	188_894	2_360_638_679	16_777_504	58_204	66_552	79_675	3_018_208_083
btreemap_rs	555_925	1_792_610_876	27_590_656	73_295	123_191	85_043	3_204_151_479
imrc_hashmap_rs	556_123	2_584_010_729	244_908_032	35_545	194_927	93_154	6_272_274_871
hashmap_rs	544_570	439_245_801	73_138_176	20_008	24_930	23_243	1_565_611_027

Priority queue

	binary_size	heapify 1m	max mem	pop_min 50	put 50	pop_min 50	upgrade
heap	166_903	5_554_617_018	24_000_360	621_690	227_224	592_588	3_189_831_485
heap_rs	540_668	139_667_629	18_284_544	55_880	21_271	55_765	648_921_342

Growable array

	binary_size	generate 5k	max mem	batch_get 500	batch_put 500	batch_remove 500	upgrade
buffer	173_903	2_601_059	65_644	95_506	803_474	173_506	3_091_310
vector	171_932	1_952_689	24_580	126_130	186_485	176_123	4_675_192
vec_rs	538_516	286_776	1_376_256	15_714	28_793	21_553	3_805_750

Stable structures

	binary_size	generate 50k	max mem	batch_get 50	batch_put 50	batch_remove 50	upgrade
btreemap_rs	555_925	76_161_521	2_555_904	62_853	95_118	84_048	139_592_121
btreemap_stable_rs	558_089	4_564_070_850	2_031_616	2_709_647	5_031_627	8_581_114	729_344
heap_rs	540_668	7_049_481	2_293_760	48_389	21_519	48_114	33_629_765
heap_stable_rs	521_160	277_865_060	458_752	2_405_453	242_829	2_387_316	729_349
vec_rs	538_516	3_077_152	2_293_760	15_714	16_643	15_941	31_301_345
vec_stable_rs	521_344	63_342_333	458_752	64_876	78_806	84_168	729_349

Environment

dfx 0.18.0

Motoko compiler 0.11.0 (source lndfxrzc-zr7pf1k6-nr3nr3d7-jfla8nbn)

rustc 1.76.0 (07dca489a 2024-02-04)

ic-repl 0.7.0

ic-wasm 0.7.0
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	193_686	267_743_355	247_834_501	33_636	24_532
Rust	541_887	82_781_801	56_787_280	41_044	39_949

Certified map

	binary_size	generate 10k	max mem	inc	witness	upgrade
Motoko	243_641	4_666_119_661	3_430_044	553_629	407_936	274_434_719
Rust	580_473	6_409_375_828	2_228_224	1_020_376	298_204	6_026_009_495

Environment

dfx 0.18.0

Motoko compiler 0.11.0 (source lndfxrzc-zr7pf1k6-nr3nr3d7-jfla8nbn)

rustc 1.76.0 (07dca489a 2024-02-04)

ic-repl 0.7.0

ic-wasm 0.7.0
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	upgrade
Motoko	273_938	510_793	22_382	18_539	19_569	157_946
Rust	847_522	506_485	88_938	117_357	112_347	1_483_364

DIP721 NFT

	binary_size	init	mint_token	transfer_token	upgrade
Motoko	220_403	481_158	30_204	8_764	89_833
Rust	879_920	203_509	303_060	71_233	1_629_856

Environment

dfx 0.18.0

Motoko compiler 0.11.0 (source lndfxrzc-zr7pf1k6-nr3nr3d7-jfla8nbn)

rustc 1.76.0 (07dca489a 2024-02-04)

ic-repl 0.7.0

ic-wasm 0.7.0
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_183	19_507
Rust	23_657	480

Timer

	binary_size	setTimer	cancelTimer
Motoko	145_619	51_778	4_626
Rust	502_754	63_379	11_676

Environment

dfx 0.18.0

Motoko compiler 0.11.0 (source lndfxrzc-zr7pf1k6-nr3nr3d7-jfla8nbn)

rustc 1.76.0 (07dca489a 2024-02-04)

ic-repl 0.7.0

ic-wasm 0.7.0
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_290	145_741	28_593	11_963	22_854	6_446
Rust	535_946	573_818	57_488	37_798	71_349	42_449

Environment

dfx 0.18.0

Motoko compiler 0.11.0 (source lndfxrzc-zr7pf1k6-nr3nr3d7-jfla8nbn)

rustc 1.76.0 (07dca489a 2024-02-04)

ic-repl 0.7.0

ic-wasm 0.7.0

dfinity / canister-profiling

bump candid #113

Map

Priority queue

Growable array

Stable structures

Statistics

SHA-2

Certified map

Statistics

Basic DAO

DIP721 NFT

Statistics

Heartbeat

Timer

Statistics

Publisher & Subscriber

Statistics

Overall Statistics

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

Publisher & Subscriber

Environment