Remove redundant syncs

[alexkorovko]

Summary by CodeRabbit

• Refactor
  • Reduced unnecessary GPU synchronization in the solver to improve runtime performance and profiling responsiveness.
  • Preserved existing behavior and public interfaces; no changes to exported APIs.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yml

Review profile: CHILL

Plan: Enterprise

Run ID: e0e40604-825f-434a-b337-4e8b3e0bd931

📥 Commits

Reviewing files that changed from the base of the PR and between 79abb83 and e391062.

📒 Files selected for processing (1)

• cunls/minimizer/gauss_newton_minimizer.cu

🚧 Files skipped from review as they are similar to previous changes (1)

• cunls/minimizer/gauss_newton_minimizer.cu

---

📝 Walkthrough

Walkthrough

Removes two explicit cudaStreamSynchronize() calls in GaussNewtonMinimizer (one in InitializeJacobian, one after the linear solve in Minimize) and reformats many function headers and blocks; runtime ordering and data movement are unchanged.

Changes

CUDA stream synchronization optimization

Layer / File(s)	Summary
Constructor, Jacobian init, and CSR conversion cunls/minimizer/gauss_newton_minimizer.cu	Constructor initializer reflowed; InitializeJacobian removes an unconditional stream sync after building triplet sparse structure and documents why it's unnecessary; CSR conversion formatting unchanged behaviour.
Cost and residual/Jacobian computation cunls/minimizer/gauss_newton_minimizer.cu	ComputeCostAsync/ComputeCost pointer/size and reduction paths reformatted; ComputeResidualAndJacobian signature/usage reflowed without logic changes.
System assembly, scaling, and mapping solution cunls/minimizer/gauss_newton_minimizer.cu	ApplyColumnScalingToNormalEquations, MapScaledLinearSolutionToTangentStep, and BuildSystem reformatted; triplet→CSR and squared matrix/RHS assembly preserved.
State updates and small helpers cunls/minimizer/gauss_newton_minimizer.cu	UpdateStates and related declarations reformatted without behavior change.
Convergence evaluation and accept/reject cunls/minimizer/gauss_newton_minimizer.cu	CheckConvergence and EvaluateAndCheckConvergence reflowed; async compute + single D2H memcpy preserved; AcceptStep/RejectStep condensed.
Minimize loop, solver init, and sync removals cunls/minimizer/gauss_newton_minimizer.cu	Minimize function formatting adjusted; solver initialization and profiling unchanged; per-iteration linear-solve block no longer calls cudaStreamSynchronize(stream) before leaving the NVTX scope; loop logging reformatted.

🎯 4 (Complex) | ⏱️ ~45 minutes

"I hopped through kernels, quiet and light,
Removing the sync that kept hops from flight.
Streams now follow their ordered way,
Jacobians build, solvers play — no delay.
🐇✨"

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'Remove redundant syncs' directly and concisely describes the main change: removing unnecessary CUDA stream synchronization calls.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch remove_syncs

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}