docs(benchmarks): record authoritative baseline C1/C2 on HDFS_v1 (§9.6)

[jensholdgaard]

What

Adds §9.6 to docs/benchmarks.md: the authoritative baseline-8vcpu-32gib C1/C2 rerun on the representative LogHub HDFS_v1 corpus (11.2 M lines) — the run §9.5 anticipated for RFC 0001's validated flip.

How

Provisioned a fresh 8-vCPU/32-GiB dedicated host, fetched HDFS_v1 on the box (md5-verified, never redistributed), ran ourios-bench --gates c1,c2 --hardware-kind baseline-8vcpu-32gib at git 9a57ace, pulled the results JSON, and tore the box down immediately after.

gate	result	verdict
C1	1.000000 — 11,175,578/11,175,578 non-lossy reconstruct bit-identical; lossy 4.6e-06 (51 rows)	PASS
C2	40 templates @ 11.2 M lines (33 @ 1 M), ratio 0.825, formal gate applies	PASS

The verdicts match §9.5's local diagnostic run bit-for-bit — expected, since C1 (reconstruction fidelity) and C2 (template-count convergence) are deterministic functions of (corpus, miner) with no wall-clock or hardware-sensitive component (contrast A1's writer-environment sensitivity, §9.4). The value of this run is the authoritative hardware_kind stamp.

Invariants (CLAUDE.md §3)

Doc-only change. No code, no schema, no hot path. §3.1 (reconstruction/lossy retention) and §3.3 (bit-identical reconstruction) are the invariants C1 measures, and the recorded reading is PASS.

Follow-up (not in this PR)

Under RFC 0011 these are RFC 0001's validated gates; both now PASS authoritatively, leaving the RFC 0001 → validated status flip to the maintainer.

🤖 Generated with Claude Code

Summary by CodeRabbit

• Documentation
  • Updated benchmarks documentation with comprehensive validation results for HDFS_v1. Results document performance metrics and baselines on standard hardware configurations, confirming successful pass criteria for bit-identical reconstruction accuracy and template count convergence benchmarks. These authoritative results establish hardware-stamped performance baselines and serve as the foundation for the validation framework.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 804b053a-3e1c-4a17-a816-348a6e43834c

📥 Commits

Reviewing files that changed from the base of the PR and between 9a57ace and 5117f04.

📒 Files selected for processing (1)

• docs/benchmarks.md

---

📝 Walkthrough

Walkthrough

docs/benchmarks.md gains a new §9.6 section recording the authoritative C1/C2 benchmark run for HDFS_v1 on baseline-8vcpu-32gib. It documents the ourios-bench --gates c1,c2 invocation at git 9a57ace, a results table with PASS verdicts for both gates, and a confirmation paragraph tying these results to RFC 0001 validated gates.

Changes

§9.6 Authoritative C1/C2 HDFS_v1 Benchmark Results

Layer / File(s)	Summary
§9.6 results table and authoritative confirmation docs/benchmarks.md	Inserts the §9.6 run descriptor and results table (C1 PASS at reconstruct rate 1.000000, C2 PASS at end template count 40/ratio 0.825) and replaces prior placeholder text with a paragraph confirming bit-for-bit match with §9.5 as the hardware-stamped basis for RFC 0001 validated gates.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

• jensholdgaard/ourios#193: Directly related — contains the preceding §9.5 HDFS_v1 A1/C1/C2 diagnostic run and RFC 0001 validated gating changes that §9.6 references and confirms bit-for-bit.
• jensholdgaard/ourios#51: Implements the C1 end-to-end reconstruction-rate measurement logic (C1Accumulator/harness) whose output is the C1 PASS verdict recorded in §9.6.
• jensholdgaard/ourios#85: Also modifies docs/benchmarks.md to record C1/C2 benchmark outcomes in an earlier section, forming the continuous benchmark documentation trail that §9.6 extends.

Poem

🐇 Hop, hop, the numbers are in,
C1 and C2 both gleam with a grin,
One-point-zero, the reconstruct shines,
Forty templates drawn in clean lines,
The hardware has stamped it — we win! ✅

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically summarizes the main change: recording authoritative C1/C2 benchmark results for HDFS_v1 on the baseline host, with precise section reference (§9.6).
Description check	✅ Passed	The description comprehensively covers the PR objectives with detailed sections on What, How, results table, invariants, and follow-up; however, it does not include the standard template's Checklist section with items like cargo fmt/clippy verification and tests/docs updates.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
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 unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch docs/benchmarks-authoritative-c1c2

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[Copilot]

Pull request overview

Adds an authoritative record of the baseline (baseline-8vcpu-32gib) C1/C2 benchmark rerun on the LogHub HDFS_v1 corpus, capturing the hardware-kind-stamped results needed for RFC 0001’s validated gate evidence.

Changes:

• Updates the §9.5 narrative to reference the completed authoritative rerun on 2026-06-14.
• Adds new §9.6 with corpus/hardware/run provenance and a C1/C2 results table for HDFS_v1.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.