Restore CRDT and CSPA throughput with memory admission enabled

Closed
  • Is something wrong?
#1,913 2 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
35/100
Issue type
Refactor
Clarity
Mostly clear
Activity status
Active
Tech stack
c
Domain
performance

Research direction

Profile the CRDT and CSPA workloads on the baseline and memory-admission candidate in the pinned performance environment, starting with the existing crdt_perf_gate and cspa_w1_gate entry points. Identify the governed allocation or admission hot path before changing it, then verify the current targets, focused denial/ownership/rollback tests, accounting, typed errors, and result correctness without weakening gates.

Written by the indexing model from the issue text.

Description

p1-high

Discovery context

The required Perf Suite for PR #1903 (issue #1369), rebased on main at 8d91c2da, failed both single-worker performance gates in run 35906306991, job 107334769925:

  • crdt_perf_gate: median 42,324.7 ms against the unchanged 38,120 ms target (CoV 0.446%).
  • cspa_w1_gate: median 3,735.1 ms against the unchanged 2,050 ms target (CoV 1.631%).

The expected CRDT result and CSPA tuple/iteration counts were correct. The CI measurements correlate with the memory-admission work in #1369, but do not identify the responsible hot path. Profile the baseline and candidate before assigning a cause; session setup or hash rebuilding alone is not yet shown to explain the CSPA regression.

Scope

Compare and profile the CRDT and CSPA workloads on the baseline and memory-admission candidate under the required pinned performance environment. Optimize measured governed allocation/admission hot paths while preserving per-class accounting, finite budget denial, typed errors, ownership transfer, rollback, and result correctness. Add or update focused tests if an optimization changes admission behavior.

Acceptance criteria

  • The existing required crdt_perf_gate passes its current 38,120 ms target and reports reproducible before/after measurements and variance.
  • The existing required cspa_w1_gate passes its current 2,050 ms target and reports reproducible before/after measurements and variance.
  • Focused memory-admission denial, ownership, and rollback tests pass for any changed paths.
  • No performance threshold or baseline is raised, and no gate is skipped or weakened.
  • Architect and Critic review the measured diagnosis and optimization before implementation.

Related: #1369, PR #1903.

Dominant language
C
Stars
18
Forks
2
Avg merge
9h 44m
Merged PRs (30d)
162

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from semantic-reasoning/wirelog

All issues in semantic-reasoning/wirelog

Similar issues

More C issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.