enhance(normalizr,endpoint): Reduce allocations in hot cache paths#3879
Merged
enhance(normalizr,endpoint): Reduce allocations in hot cache paths#3879
Conversation
Inline getCacheKey in GlobalCache to avoid eagerly creating both localCache and cycleCache Maps per entity type. Replace push(...spread) with an indexed for-loop when copying cached entity dependencies. Pre-create _removeSchema in Collection.CreateMover so normalizeMove no longer calls Object.create on every invocation, eliminating hidden class polymorphism that caused V8 "wrong call target" deoptimizations. Made-with: Cursor
|
|
The latest updates on your projects. Learn more about Vercel for GitHub. |
Contributor
|
Size Change: +29 B (+0.04%) Total Size: 80.6 kB 📦 View Changed
ℹ️ View Unchanged
|
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Benchmark React
Details
| Benchmark suite | Current: 2a50525 | Previous: 98a7831 | Ratio |
|---|---|---|---|
data-client: getlist-100 |
177.01 ops/s (± 4.6%) |
177.01 ops/s (± 5.0%) |
1 |
data-client: getlist-500 |
46.51 ops/s (± 5.7%) |
48.47 ops/s (± 5.3%) |
1.04 |
data-client: update-entity |
476.19 ops/s (± 4.6%) |
416.67 ops/s (± 6.3%) |
0.88 |
data-client: update-user |
454.55 ops/s (± 5.6%) |
434.78 ops/s (± 5.7%) |
0.96 |
data-client: getlist-500-sorted |
51.28 ops/s (± 5.1%) |
50.13 ops/s (± 7.4%) |
0.98 |
data-client: update-entity-sorted |
392.31 ops/s (± 6.5%) |
384.62 ops/s (± 3.2%) |
0.98 |
data-client: update-entity-multi-view |
384.62 ops/s (± 8.4%) |
322.58 ops/s (± 6.0%) |
0.84 |
data-client: list-detail-switch-10 |
12.45 ops/s (± 8.1%) |
11.87 ops/s (± 6.3%) |
0.95 |
data-client: update-user-10000 |
101.02 ops/s (± 3.5%) |
99.01 ops/s (± 4.8%) |
0.98 |
data-client: invalidate-and-resolve |
51.28 ops/s (± 4.8%) |
50.76 ops/s (± 3.7%) |
0.99 |
data-client: unshift-item |
322.58 ops/s (± 0.0%) |
333.33 ops/s (± 6.3%) |
1.03 |
data-client: delete-item |
454.55 ops/s (± 6.0%) |
416.67 ops/s (± 0.0%) |
0.92 |
data-client: move-item |
263.16 ops/s (± 3.0%) |
227.27 ops/s (± 3.3%) |
0.86 |
This comment was automatically generated by workflow using github-action-benchmark.
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #3879 +/- ##
=======================================
Coverage 98.10% 98.11%
=======================================
Files 153 153
Lines 2906 2913 +7
Branches 565 565
=======================================
+ Hits 2851 2858 +7
Misses 11 11
Partials 44 44 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Contributor
There was a problem hiding this comment.
Benchmark
Details
| Benchmark suite | Current: 2a50525 | Previous: 98a7831 | Ratio |
|---|---|---|---|
normalizeLong |
448 ops/sec (±1.23%) |
445 ops/sec (±2.41%) |
0.99 |
normalizeLong Values |
416 ops/sec (±0.16%) |
415 ops/sec (±0.25%) |
1.00 |
denormalizeLong |
301 ops/sec (±2.75%) |
300 ops/sec (±2.18%) |
1.00 |
denormalizeLong Values |
269 ops/sec (±2.32%) |
274 ops/sec (±2.20%) |
1.02 |
denormalizeLong donotcache |
1021 ops/sec (±0.14%) |
1027 ops/sec (±0.40%) |
1.01 |
denormalizeLong Values donotcache |
760 ops/sec (±0.18%) |
761 ops/sec (±0.31%) |
1.00 |
denormalizeShort donotcache 500x |
1570 ops/sec (±0.64%) |
1584 ops/sec (±0.61%) |
1.01 |
denormalizeShort 500x |
849 ops/sec (±2.24%) |
864 ops/sec (±2.10%) |
1.02 |
denormalizeShort 500x withCache |
7526 ops/sec (±0.13%) |
6266 ops/sec (±0.11%) |
0.83 |
queryShort 500x withCache |
2969 ops/sec (±0.11%) |
2799 ops/sec (±0.19%) |
0.94 |
buildQueryKey All |
55104 ops/sec (±0.21%) |
55711 ops/sec (±0.17%) |
1.01 |
query All withCache |
6033 ops/sec (±0.19%) |
6132 ops/sec (±0.30%) |
1.02 |
denormalizeLong with mixin Entity |
284 ops/sec (±2.35%) |
285 ops/sec (±2.01%) |
1.00 |
denormalizeLong withCache |
8158 ops/sec (±0.20%) |
7880 ops/sec (±0.25%) |
0.97 |
denormalizeLong Values withCache |
5101 ops/sec (±0.11%) |
5100 ops/sec (±0.12%) |
1.00 |
denormalizeLong All withCache |
5844 ops/sec (±0.13%) |
5871 ops/sec (±0.67%) |
1.00 |
denormalizeLong Query-sorted withCache |
6057 ops/sec (±0.18%) |
6159 ops/sec (±0.26%) |
1.02 |
denormalizeLongAndShort withEntityCacheOnly |
1864 ops/sec (±0.19%) |
1786 ops/sec (±0.23%) |
0.96 |
denormalize bidirectional 50 |
5808 ops/sec (±1.85%) |
5943 ops/sec (±1.89%) |
1.02 |
denormalize bidirectional 50 donotcache |
40596 ops/sec (±0.97%) |
40836 ops/sec (±0.64%) |
1.01 |
getResponse |
4576 ops/sec (±0.60%) |
4791 ops/sec (±0.53%) |
1.05 |
getResponse (null) |
11060843 ops/sec (±0.94%) |
10513376 ops/sec (±0.93%) |
0.95 |
getResponse (clear cache) |
267 ops/sec (±1.89%) |
274 ops/sec (±1.87%) |
1.03 |
getSmallResponse |
3619 ops/sec (±0.08%) |
3370 ops/sec (±0.07%) |
0.93 |
getSmallInferredResponse |
2763 ops/sec (±0.08%) |
2572 ops/sec (±0.11%) |
0.93 |
getResponse Collection |
4464 ops/sec (±0.35%) |
4588 ops/sec (±0.42%) |
1.03 |
get Collection |
4583 ops/sec (±0.28%) |
4708 ops/sec (±0.26%) |
1.03 |
get Query-sorted |
5202 ops/sec (±0.15%) |
5380 ops/sec (±0.25%) |
1.03 |
setLong |
448 ops/sec (±0.41%) |
462 ops/sec (±0.16%) |
1.03 |
setLongWithMerge |
249 ops/sec (±0.49%) |
257 ops/sec (±0.21%) |
1.03 |
setLongWithSimpleMerge |
271 ops/sec (±0.23%) |
274 ops/sec (±0.15%) |
1.01 |
setSmallResponse 500x |
928 ops/sec (±0.14%) |
940 ops/sec (±0.10%) |
1.01 |
This comment was automatically generated by workflow using github-action-benchmark.
Made-with: Cursor
Made-with: Cursor
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
getCacheKeyinGlobalCache.getEntity: Avoids eagerly creating bothlocalCacheandcycleCacheMaps per entity type on every access. ThecycleCacheis now only created when actually entering the denormalize path (lazy initialization). Also replacespush(...spread)with an indexedfor-loop when copying cached entity dependencies to avoid spread operator overhead._removeSchemainCollection.CreateMover: Eliminates per-callObject.createinnormalizeMove, removing a source of hidden class polymorphism that caused V8 "wrong call target" deoptimizations. TheaddSchemais also eliminated sincethis.mergeis already theaddMergefunction.These optimizations target the
move-itemandupdate-user-10000React benchmark scenarios identified via--trace-opt --trace-deoptprofiling.Test plan
move-itemandupdate-user-10000React benchmarks show no regressionMade with Cursor
Note
Medium Risk
Moderate risk: changes touch core normalization/denormalization cache behavior and collection membership updates; intended semantics should be preserved but regressions would be subtle and performance-focused.
Overview
Reduces allocations in hot normalization/denormalization paths.
In
Collection.CreateMover/normalizeMove, pre-creates and reuses a_removeSchemaand stops creating per-calladdSchema/removeSchema, reusingthisfor add merges while still using the prebuilt remove schema.In
GlobalCache.getEntity, lazily initializes the per-entity-typecycleCacheonly when entering the compute/denormalize branch, and replacesdependencies.push(...deps)with an indexed loop to avoid spread overhead.Reviewed by Cursor Bugbot for commit 2a50525. Bugbot is set up for automated code reviews on this repo. Configure here.