feat: CPU utilization optimization in image diff calculations#146
Merged
abelonogov-ld merged 48 commits intomainfrom Mar 6, 2026
Merged
feat: CPU utilization optimization in image diff calculations#146abelonogov-ld merged 48 commits intomainfrom
abelonogov-ld merged 48 commits intomainfrom
Conversation
(cherry picked from commit 78ad485)
…anvas buffer limit. Updated makeExportFrame to accept keyFrameId and isKeyframe parameters. Adjusted canvasBufferLimit to account for keyframe logic. Added test to verify mutation appending when canvas buffer limit is exceeded for non-keyframes.
- Added support for command-line options including scheme, destination, only-testing, skip-testing, and test-file. - Introduced a usage function to display help information. - Implemented a function to derive test identifiers from file paths. - Refactored the test execution command to accommodate new options and improve readability. (cherry picked from commit cd1e1e2)
(cherry picked from commit 54853ad)
… allowing multiple runs. Updated BenchmarkExecutor to return execution time alongside bytes, and modified BenchmarkView to display execution time in results.
- Updated the ExportFrame structure to replace TileSignature with ImageSignature in AddImage and RemoveImage. - Adjusted RRWebEventGenerator and related methods to accommodate the new image signature handling. - Modified unit tests to ensure compatibility with the updated ExportFrame parameters.
(cherry picked from commit 37b2dfd1eb49b3e6f0e64441523119dc2a9111fb)
(cherry picked from commit 8ea1c7b)
…ash-swiftpackage * andrey/exportdiffmanager: separated time Alert errors do account disk reading into benchmark # Conflicts: # Sources/LaunchDarklySessionReplay/BenchMark/BenchmarkExecutor.swift
* main: chore(main): release 0.22.0 (#145) feat: Moves compression before buffer (Optimization) (#144) chore(main): release 0.21.0 (#143) feat: Bridging hooks calls from MAUI (#142) # Conflicts: # Sources/LaunchDarklySessionReplay/BenchMark/BenchmarkExecutor.swift # Sources/LaunchDarklySessionReplay/ScreenCapture/TileSignatureManager.swift # TestApp/Sources/BenchmarkView.swift
abelonogov-ld
requested review from
a team and
mario-launchdarkly
as code owners
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
Vadman97
approved these changes
Mar 5, 2026
mario-launchdarkly
approved these changes
Mar 6, 2026
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.
| result.hashHi = (int64_t)h1; | ||
| return result; | ||
| } | ||
| #endif |
There was a problem hiding this comment.
NEON function guard mismatch causes debug build linker error
High Severity
The tile_hash_w64_neon function is declared in the header under #if defined(__ARM_NEON) but its implementation is guarded by #if USE_NEON, which requires both __ARM_NEON and __OPTIMIZE__. In debug builds on ARM (devices and Apple Silicon simulators), __OPTIMIZE__ is not defined, so the declaration is visible but the symbol is never compiled. The parity tests in TileHashParityTests.swift call tile_hash_w64_neon unconditionally, resulting in a linker error in any non-optimized ARM build.
Additional Locations (1)
abelonogov-ld
pushed a commit
that referenced
this pull request
Mar 7, 2026
🤖 I have created a release *beep* *boop* --- ## [0.23.0](0.22.0...0.23.0) (2026-03-06) ### Features * CPU utilization optimization in image diff calculations ([#146](#146)) ([80e0cf8](80e0cf8)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please). <!-- CURSOR_SUMMARY --> --- > [!NOTE] > **Low Risk** > Low risk: this PR only updates version numbers and release notes (no runtime code changes beyond `sdkVersion` constant). > > **Overview** > Releases version `0.23.0` by updating the release manifest, CocoaPods podspec versions, and the exported `sdkVersion` constant. > > Updates `CHANGELOG.md` with the `0.23.0` entry noting a CPU utilization optimization in image diff calculations. > > <sup>Written by [Cursor Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit db4a44b. This will update automatically on new commits. Configure [here](https://cursor.com/dashboard?tab=bugbot).</sup> <!-- /CURSOR_SUMMARY -->
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
3 participants
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.
Note
High Risk
Replaces a core image-diff primitive with new low-level C/NEON code and unsafe buffer plumbing, which could affect correctness or stability of session replay capture if edge cases slip through.
Overview
Major perf change: Tile signature generation used for image diffs is rewritten to use a new
SessionReplayCtarget (C + optional ARM NEON), replacing the prior Swift/CommonCrypto SHA-256 per-tile hashing with a faster non-crypto 128-bit style hash and centralized tile layout computation.TileSignatureManagernow callstile_compute_layout/tile_compute_alland reuses an internal C buffer;ImageSignatureis split into its own file and now precomputes an accumulated hash to speed dictionary lookups/equality.Validation/benchmarking: Adds
TileHashParityTeststo ensure NEON and scalar paths match, updates the benchmark tooling/UI to measure signature computation, and forces decoded frame images inRawFrameReaderfor more consistent pixel access. Packaging is updated so CocoaPods and SPM both build/link the newSessionReplayCmodule.Written by Cursor Bugbot for commit d3604f4. This will update automatically on new commits. Configure here.