Skip to content

perf: fast-path small candid numbers - #709

Merged
lwshang merged 6 commits into
masterfrom
perf-1-fast-path-small-num
Mar 15, 2026
Merged

perf: fast-path small candid numbers#709
lwshang merged 6 commits into
masterfrom
perf-1-fast-path-small-num

Conversation

@sasa-tomic

@sasa-tomicsasa-tomic commented Mar 13, 2026•
edited
Loading

Copy link
Copy Markdown
Contributor

Overview
Performance improvement

Requirements
Preserve existing wire compatibility and keep record decoding behavior unchanged.

Solution
Encode and decode Nat and Int values through native LEB128 when they fit in machine integers so common small numbers avoid bigint work.

Considerations
I expect performance improvement and full forward and backward compatibility

Ref. #710

@sasa-tomic
perf: fast-path small candid numbers
f059164
Encode and decode Nat and Int values through native LEB128 when they fit in machine integers so common small numbers avoid bigint work.
@github-actions

github-actionsBot commented Mar 13, 2026•
edited
Loading

Copy link
Copy Markdown
NameMax Mem (Kb)EncodeDecode
blob4_2244_207_487 ($\textcolor{green}{-0.01\%}$)2_122_432 ($\textcolor{green}{-0.00\%}$)
btreemap73_856 ($\textcolor{green}{-2.12\%}$)531_975_943 ($\textcolor{green}{-88.77\%}$)13_040_266_452 ($\textcolor{green}{-14.16\%}$)
nns1922_021_253 ($\textcolor{green}{-0.27\%}$)5_714_511 ($\textcolor{green}{-0.55\%}$)
nns_list_proposal1_2167_015_064 ($\textcolor{green}{-0.47\%}$)67_819_354 ($\textcolor{green}{-0.07\%}$)
option_list128716_025 ($\textcolor{green}{-91.14\%}$)23_547_997 ($\textcolor{green}{-11.30\%}$)
text6_3364_204_384 ($\textcolor{green}{-0.01\%}$)7_877_759 ($\textcolor{green}{-0.00\%}$)
variant_list128711_051 ($\textcolor{green}{-91.26\%}$)22_235_339 ($\textcolor{green}{-11.67\%}$)
vec_int1616_704123_694_298 ($\textcolor{green}{-0.00\%}$)1_015_045_667 ($\textcolor{green}{-0.00\%}$)
  • Parser cost: 17_069_949
  • Extra args: 3_416_403 ($\textcolor{red}{0.00\%}$)
Click to see raw report
---------------------------------------------------
Benchmark: blob
total:
instructions: 6.33 M (-0.00%) (change within noise threshold)
heap_increase: 66 pages (no change)
stable_memory_increase: 0 pages (no change)
1. Encoding (scope):
calls: 1 (no change)
instructions: 4.21 M (-0.01%) (change within noise threshold)
heap_increase: 66 pages (no change)
stable_memory_increase: 0 pages (no change)
2. Decoding (scope):
calls: 1 (no change)
instructions: 2.12 M (-0.00%) (change within noise threshold)
heap_increase: 0 pages (no change)
stable_memory_increase: 0 pages (no change)
---------------------------------------------------
Benchmark: btreemap
total:
instructions: 13.57 B (improved by 31.89%)
heap_increase: 1154 pages (improved by 2.12%)
stable_memory_increase: 0 pages (no change)
1. Encoding (scope):
calls: 1 (no change)
instructions: 531.98 M (improved by 88.77%)
heap_increase: 159 pages (no change)
stable_memory_increase: 0 pages (no change)
2. Decoding (scope):
calls: 1 (no change)
instructions: 13.04 B (improved by 14.16%)
heap_increase: 995 pages (improved by 2.45%)
stable_memory_increase: 0 pages (no change)
---------------------------------------------------
Benchmark: extra_args
total:
instructions: 3.42 M (0.00%) (change within noise threshold)
heap_increase: 0 pages (no change)
stable_memory_increase: 0 pages (no change)
---------------------------------------------------
Benchmark: nns
total:
instructions: 25.65 M (-0.14%) (change within noise threshold)
heap_increase: 3 pages (no change)
stable_memory_increase: 0 pages (no change)
0. Parsing (scope):
calls: 1 (no change)
instructions: 17.07 M (no change)
heap_increase: 3 pages (no change)
stable_memory_increase: 0 pages (no change)
1. Encoding (scope):
calls: 1 (no change)
instructions: 2.02 M (-0.27%) (change within noise threshold)
heap_increase: 0 pages (no change)
stable_memory_increase: 0 pages (no change)
2. Decoding (scope):
calls: 1 (no change)
instructions: 5.71 M (-0.55%) (change within noise threshold)
heap_increase: 0 pages (no change)
stable_memory_increase: 0 pages (no change)
---------------------------------------------------
Benchmark: nns_list_proposal
total:
instructions: 74.84 M (-0.11%) (change within noise threshold)
heap_increase: 19 pages (no change)
stable_memory_increase: 0 pages (no change)
1. Encoding (scope):
calls: 1 (no change)
instructions: 7.02 M (-0.47%) (change within noise threshold)
heap_increase: 5 pages (no change)
stable_memory_increase: 0 pages (no change)
2. Decoding (scope):
calls: 1 (no change)
instructions: 67.82 M (-0.07%) (change within noise threshold)
heap_increase: 14 pages (no change)
stable_memory_increase: 0 pages (no change)
---------------------------------------------------
Benchmark: option_list
total:
instructions: 24.27 M (improved by 29.93%)
heap_increase: 2 pages (no change)
stable_memory_increase: 0 pages (no change)
1. Encoding (scope):
calls: 1 (no change)
instructions: 716.02 K (improved by 91.14%)
heap_increase: 0 pages (no change)
stable_memory_increase: 0 pages (no change)
2. Decoding (scope):
calls: 1 (no change)
instructions: 23.55 M (improved by 11.30%)
heap_increase: 2 pages (no change)
stable_memory_increase: 0 pages (no change)
---------------------------------------------------
Benchmark: text
total:
instructions: 12.08 M (-0.00%) (change within noise threshold)
heap_increase: 99 pages (no change)
stable_memory_increase: 0 pages (no change)
1. Encoding (scope):
calls: 1 (no change)
instructions: 4.20 M (-0.01%) (change within noise threshold)
heap_increase: 66 pages (no change)
stable_memory_increase: 0 pages (no change)
2. Decoding (scope):
calls: 1 (no change)
instructions: 7.88 M (-0.00%) (change within noise threshold)
heap_increase: 33 pages (no change)
stable_memory_increase: 0 pages (no change)
---------------------------------------------------
Benchmark: variant_list
total:
instructions: 22.95 M (improved by 31.10%)
heap_increase: 2 pages (no change)
stable_memory_increase: 0 pages (no change)
1. Encoding (scope):
calls: 1 (no change)
instructions: 711.05 K (improved by 91.26%)
heap_increase: 0 pages (no change)
stable_memory_increase: 0 pages (no change)
2. Decoding (scope):
calls: 1 (no change)
instructions: 22.24 M (improved by 11.67%)
heap_increase: 2 pages (no change)
stable_memory_increase: 0 pages (no change)
---------------------------------------------------
Benchmark: vec_int16
total:
instructions: 1.14 B (-0.00%) (change within noise threshold)
heap_increase: 261 pages (no change)
stable_memory_increase: 0 pages (no change)
1. Encoding (scope):
calls: 1 (no change)
instructions: 123.69 M (-0.00%) (change within noise threshold)
heap_increase: 261 pages (no change)
stable_memory_increase: 0 pages (no change)
2. Decoding (scope):
calls: 1 (no change)
instructions: 1.02 B (-0.00%) (change within noise threshold)
heap_increase: 0 pages (no change)
stable_memory_increase: 0 pages (no change)
---------------------------------------------------
Summary:
instructions:
status: Improvements detected 🟢
counts: [total 9 | regressed 0 | improved 3 | new 0 | unchanged 6]
change: [max +102 | p75 -288 | median -37.23K | p25 -10.36M | min -6.35B]
change %: [max 0.00% | p75 -0.00% | median -0.11% | p25 -29.93% | min -31.89%]
heap_increase:
status: Improvements detected 🟢
counts: [total 9 | regressed 0 | improved 1 | new 0 | unchanged 8]
change: [max 0 | p75 0 | median 0 | p25 0 | min -25]
change %: [max 0.00% | p75 0.00% | median 0.00% | p25 0.00% | min -2.12%]
stable_memory_increase:
status: No significant changes 👍
counts: [total 9 | regressed 0 | improved 0 | new 0 | unchanged 9]
change: [max 0 | p75 0 | median 0 | p25 0 | min 0]
change %: [max 0.00% | p75 0.00% | median 0.00% | p25 0.00% | min 0.00%]
---------------------------------------------------
Only significant changes:
| status | name | calls | ins | ins Δ% | HI | HI Δ% | SMI | SMI Δ% |
|--------|---------------------------|-------|---------|---------|-------|--------|-----|---------|
| - | option_list::2. Decoding | 1 | 23.55M | -11.30% | 2 | 0.00% | 0 | 0.00% |
| - | variant_list::2. Decoding | 1 | 22.24M | -11.67% | 2 | 0.00% | 0 | 0.00% |
| - | btreemap::2. Decoding | 1 | 13.04B | -14.16% | 995 | -2.45% | 0 | 0.00% |
| - | option_list | | 24.27M | -29.93% | 2 | 0.00% | 0 | 0.00% |
| - | variant_list | | 22.95M | -31.10% | 2 | 0.00% | 0 | 0.00% |
| - | btreemap | | 13.57B | -31.89% | 1.15K | -2.12% | 0 | 0.00% |
| - | btreemap::1. Encoding | 1 | 531.98M | -88.77% | 159 | 0.00% | 0 | 0.00% |
| - | option_list::1. Encoding | 1 | 716.02K | -91.14% | 0 | 0.00% | 0 | 0.00% |
| - | variant_list::1. Encoding | 1 | 711.05K | -91.26% | 0 | 0.00% | 0 | 0.00% |
ins = instructions, HI = heap_increase, SMI = stable_memory_increase, Δ% = percent change
---------------------------------------------------
Successfully persisted results to canbench_results.yml

@sasa-tomic
sasa-tomic marked this pull request as ready for review March 13, 2026 09:36
@sasa-tomic
sasa-tomic requested a review from a team as a code ownerMarch 13, 2026 09:36
lwshangand others added 4 commits March 15, 2026 12:47
@lwshang@claude
perf: add fast-path for Int::decode to match Nat::decode
5f4f2d7
Mirror the u64-accumulation fast path already added to Nat::decode.
Accumulate bits into a u64 with the same overflow guard, then
sign-extend at the last byte to produce an i64. Values that fit in
i64 skip BigInt allocation entirely; values in [2^63, 2^64-1] (valid
Int but not i64) still use BigInt correctly. The BigInt fallback for
truly large numbers mirrors the structure of Nat::decode's fallback.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
@lwshang@claude
test: add boundary tests for Nat/Int decode fast-path transitions
76060b0
Cover the exact values where the u64 fast path hands off to BigInt:
- Nat: u64::MAX (last fast-path value) and u64::MAX+1 (first BigInt),
plus a sweep of values near the boundary.
- Int: i64::MAX and i64::MAX+1 (positive overflow branch), i64::MIN
(sign-extension fast path) and i64::MIN-1 (BigInt negative branch).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
@lwshang@claude
docs: clarify fits_u64 overflow guard in Nat::decode and Int::decode
8452ab2
Add inline comments explaining why the `shift < 64` branch must be
tested before evaluating `1u64 << (64 - shift)`: the expression would
panic in debug mode if shift were >= 64, so the else-false arm acts as
both the panic guard and the BigInt-fallback trigger.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
@lwshang@claude
refactor: extract leb128_fits_u64 helper to deduplicate overflow check
1aa734d
The identical 7-line fits_u64 guard existed verbatim in both Nat::decode
and Int::decode. Extract it into a shared inline helper function.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
@lwshang
lwshang merged commit 7319172 into masterMar 15, 2026
11 checks passed
@lwshang
lwshang deleted the perf-1-fast-path-small-num branch March 15, 2026 17:00
Sign up for freeto join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@lwshanglwshanglwshang approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@sasa-tomic@lwshang