Profile wasm-opt level 2

[github-actions[bot]]

Warning The flamegraph link only works after you merge.

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	121_920	11_354
Rust	27_253	1_080

Timer

	binary_size	setTimer	cancelTimer
Motoko	135_637	32_210	1_748
Rust	424_735	53_825	10_153

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 50k. Insert 50k Nat32 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_live_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 nlogn) 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.
• hashmap_rs uses the fxhash crate, which is the same as std::collections::HashMap, but with a deterministic hasher. This ensures reproducible result.
• btree comes from Byron Becker's stable BTreeMap library.
• zhenya_hashmap comes from Zhenya Usenko's stable HashMap library.
• The MoVM table measures the performance of an experimental implementation of Motoko interpreter. External developers can ignore this table for now.

Map

	binary_size	generate 50k	max mem	batch_get 50	batch_put 50	batch_remove 50
hashmap	162_667	2_112_036_256	9_102_052	1_124_437	613_910_550	1_065_165
triemap	166_893	2_036_258_951	9_716_008	781_608	1_868_829	1_042_928
rbtree	165_811	1_895_737_241	10_102_184	709_153	1_705_707	946_478
splay	163_998	2_105_962_355	9_302_108	1_155_004	1_976_076	1_155_170
btree	194_059	1_934_621_105	8_157_968	883_615	1_773_826	956_678
zhenya_hashmap	157_997	1_648_201_768	9_301_800	649_931	1_454_771	654_626
btreemap_rs	422_180	122_486_096	1_638_400	60_360	138_077	61_505
hashmap_rs	411_353	52_292_656	1_835_008	20_648	62_156	21_930

Priority queue

	binary_size	heapify 50k	mem	pop_min 50	put 50
heap	150_084	695_512_639	1_400_024	377_593	727_475	379_121
heap_rs	390_095	4_978_816	819_200	52_862	21_741	53_054

MoVM

	binary_size	generate 10k	max mem	batch_get 50	batch_put 50	batch_remove 50
hashmap	162_667	422_411_986	1_820_844	1_122_416	123_656_843	1_062_614
hashmap_rs	411_353	10_772_672	950_272	19_979	61_484	20_872
imrc_hashmap_rs	420_546	19_678_328	1_572_864	31_170	118_660	37_893
movm_rs	1_704_006	1_097_944_086	2_654_208	2_716_088	6_930_281	5_411_168
movm_dynamic_rs	1_868_114	551_513_510	2_129_920	2_170_192	2_980_021	2_146_546

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	142_143	129_663	18_768	8_564	14_734	3_720
Rust	459_879	513_055	61_293	41_686	86_682	48_938

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	232_614	41_548	18_372	12_873	15_115
Rust	734_143	526_210	99_189	122_212	134_838

DIP721 NFT

	binary_size	init	mint_token	transfer_token
Motoko	185_377	12_292	23_080	4_818
Rust	775_133	142_951	370_840	92_651

Motoko Garbage Collection

Measure Motoko garbage collection cost using the Triemap benchmark. The max mem column reports rts_max_live_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.

• 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.

	generate 80k	max mem	batch_get 50	batch_put 50	batch_remove 50
default	247_115_881	15_539_984	50	50	50
copying	247_115_831	15_539_984	247_110_319	247_262_382	247_262_501
compacting	409_365_425	15_539_984	308_339_012	348_775_445	352_663_118
generational	624_423_107	15_540_260	57_009	1_390_483	1_060_163

[github-actions[bot]]

Note Diffing the performance result against the published result from main branch

Heartbeat

	binary_size	heartbeat
Motoko	121_920 ($\textcolor{green}{-17.13\%}$)	11_354 ($\textcolor{green}{-4.89\%}$)
Rust	27_253 ($\textcolor{green}{-23.55\%}$)	1_080 ($\textcolor{red}{83.99\%}$)

Timer

	binary_size	setTimer	cancelTimer
Motoko	135_637 ($\textcolor{green}{-16.76\%}$)	32_210 ($\textcolor{green}{-7.01\%}$)	1_748 ($\textcolor{green}{-9.10\%}$)
Rust	424_735 ($\textcolor{green}{-19.20\%}$)	53_825 ($\textcolor{green}{-3.55\%}$)	10_153 ($\textcolor{green}{-3.68\%}$)

Map

	binary_size	generate 50k	max mem	batch_get 50	batch_put 50	batch_remove 50
hashmap	162_667 ($\textcolor{green}{-16.84\%}$)	2_112_036_256 ($\textcolor{green}{-11.52\%}$)	9_102_052	1_124_437 ($\textcolor{green}{-13.06\%}$)	613_910_550 ($\textcolor{green}{-10.92\%}$)	1_065_165 ($\textcolor{green}{-13.06\%}$)
triemap	166_893 ($\textcolor{green}{-17.13\%}$)	2_036_258_951 ($\textcolor{green}{-11.06\%}$)	9_716_008	781_608 ($\textcolor{green}{-12.48\%}$)	1_868_829 ($\textcolor{green}{-11.65\%}$)	1_042_928 ($\textcolor{green}{-12.47\%}$)
rbtree	165_811 ($\textcolor{green}{-16.92\%}$)	1_895_737_241 ($\textcolor{green}{-10.48\%}$)	10_102_184	709_153 ($\textcolor{green}{-13.90\%}$)	1_705_707 ($\textcolor{green}{-11.04\%}$)	946_478 ($\textcolor{green}{-12.52\%}$)
splay	163_998 ($\textcolor{green}{-17.02\%}$)	2_105_962_355 ($\textcolor{green}{-10.76\%}$)	9_302_108	1_155_004 ($\textcolor{green}{-11.53\%}$)	1_976_076 ($\textcolor{green}{-11.21\%}$)	1_155_170 ($\textcolor{green}{-11.58\%}$)
btree	194_059 ($\textcolor{green}{-17.52\%}$)	1_934_621_105 ($\textcolor{green}{-10.84\%}$)	8_157_968	883_615 ($\textcolor{green}{-12.24\%}$)	1_773_826 ($\textcolor{green}{-11.31\%}$)	956_678 ($\textcolor{green}{-12.19\%}$)
zhenya_hashmap	157_997 ($\textcolor{green}{-16.46\%}$)	1_648_201_768 ($\textcolor{green}{-11.16\%}$)	9_301_800	649_931 ($\textcolor{green}{-12.91\%}$)	1_454_771 ($\textcolor{green}{-11.92\%}$)	654_626 ($\textcolor{green}{-13.02\%}$)
btreemap_rs	422_180 ($\textcolor{green}{-18.20\%}$)	122_486_096 ($\textcolor{green}{-1.06\%}$)	1_638_400	60_360 ($\textcolor{red}{1.07\%}$)	138_077 ($\textcolor{green}{-1.56\%}$)	61_505 ($\textcolor{green}{-0.94\%}$)
hashmap_rs	411_353 ($\textcolor{green}{-18.41\%}$)	52_292_656 ($\textcolor{green}{-1.77\%}$)	1_835_008	20_648 ($\textcolor{green}{-3.34\%}$)	62_156 ($\textcolor{green}{-2.57\%}$)	21_930 ($\textcolor{green}{-3.72\%}$)

Priority queue

	binary_size	heapify 50k	mem	pop_min 50	put 50
heap	150_084 ($\textcolor{green}{-16.80\%}$)	695_512_639 ($\textcolor{green}{-12.67\%}$)	1_400_024	377_593 ($\textcolor{green}{-10.27\%}$)	727_475 ($\textcolor{green}{-12.82\%}$)
heap_rs	390_095 ($\textcolor{green}{-17.90\%}$)	4_978_816 ($\textcolor{green}{-1.25\%}$)	819_200	52_862 ($\textcolor{green}{-1.31\%}$)	21_741 ($\textcolor{green}{-2.42\%}$)

MoVM

	binary_size	generate 10k	max mem	batch_get 50	batch_put 50	batch_remove 50
hashmap	162_667 ($\textcolor{green}{-16.84\%}$)	422_411_986 ($\textcolor{green}{-11.53\%}$)	1_820_844	1_122_416 ($\textcolor{green}{-13.09\%}$)	123_656_843 ($\textcolor{green}{-10.96\%}$)	1_062_614 ($\textcolor{green}{-13.08\%}$)
hashmap_rs	411_353 ($\textcolor{green}{-18.41\%}$)	10_772_672 ($\textcolor{green}{-1.75\%}$)	950_272	19_979 ($\textcolor{green}{-3.37\%}$)	61_484 ($\textcolor{green}{-2.56\%}$)	20_872 ($\textcolor{green}{-3.67\%}$)
imrc_hashmap_rs	420_546 ($\textcolor{green}{-18.59\%}$)	19_678_328 ($\textcolor{green}{-0.92\%}$)	1_572_864	31_170 ($\textcolor{green}{-2.04\%}$)	118_660 ($\textcolor{green}{-1.29\%}$)	37_893 ($\textcolor{green}{-0.07\%}$)
movm_rs	1_704_006 ($\textcolor{green}{-16.27\%}$)	1_097_944_086 ($\textcolor{green}{-0.08\%}$)	2_654_208	2_716_088 ($\textcolor{green}{-1.02\%}$)	6_930_281 ($\textcolor{green}{-0.19\%}$)	5_411_168 ($\textcolor{green}{-0.10\%}$)
movm_dynamic_rs	1_868_114 ($\textcolor{green}{-17.04\%}$)	551_513_510 ($\textcolor{green}{-0.73\%}$)	2_129_920	2_170_192 ($\textcolor{green}{-0.76\%}$)	2_980_021 ($\textcolor{green}{-1.00\%}$)	2_146_546 ($\textcolor{green}{-0.90\%}$)

Publisher & Subscriber

	pub_binary_size	sub_binary_size	subscribe_caller	subscribe_callee	publish_caller	publish_callee
Motoko	142_143 ($\textcolor{green}{-17.24\%}$)	129_663 ($\textcolor{green}{-17.36\%}$)	18_768 ($\textcolor{green}{-4.45\%}$)	8_564 ($\textcolor{green}{-6.35\%}$)	14_734 ($\textcolor{green}{-5.16\%}$)	3_720 ($\textcolor{green}{-7.02\%}$)
Rust	459_879 ($\textcolor{green}{-18.46\%}$)	513_055 ($\textcolor{green}{-26.31\%}$)	61_293 ($\textcolor{green}{-3.33\%}$)	41_686 ($\textcolor{green}{-3.48\%}$)	86_682 ($\textcolor{green}{-3.02\%}$)	48_938 ($\textcolor{green}{-3.18\%}$)

Basic DAO

	binary_size	init	transfer_token	submit_proposal	vote_proposal
Motoko	232_614 ($\textcolor{green}{-20.08\%}$)	41_548 ($\textcolor{green}{-6.94\%}$)	18_372 ($\textcolor{green}{-8.50\%}$)	12_873 ($\textcolor{green}{-9.09\%}$)	15_115 ($\textcolor{green}{-10.15\%}$)
Rust	734_143 ($\textcolor{green}{-22.27\%}$)	526_210 ($\textcolor{green}{-2.91\%}$)	99_189 ($\textcolor{green}{-3.20\%}$)	122_212 ($\textcolor{green}{-2.91\%}$)	134_838 ($\textcolor{green}{-2.86\%}$)

DIP721 NFT

	binary_size	init	mint_token	transfer_token
Motoko	185_377 ($\textcolor{green}{-21.20\%}$)	12_292 ($\textcolor{green}{-8.12\%}$)	23_080 ($\textcolor{green}{-6.48\%}$)	4_818 ($\textcolor{green}{-10.06\%}$)
Rust	775_133 ($\textcolor{green}{-22.34\%}$)	142_951 ($\textcolor{green}{-2.85\%}$)	370_840 ($\textcolor{green}{-2.72\%}$)	92_651 ($\textcolor{green}{-2.95\%}$)

Motoko Garbage Collection

	generate 80k	max mem	batch_get 50	batch_put 50	batch_remove 50
default	247_115_881	15_539_984	50	50	50
copying	247_115_831	15_539_984	247_110_319	247_262_382	247_262_501
compacting	409_365_425	15_539_984	308_339_012	348_775_445	352_663_118
generational	624_423_107	15_540_260	57_009	1_390_483	1_060_163