chore(bench): publish proxy and memory HNSW and ranking-quality and write-pipeline benchmark data - #683
Conversation
Signed-off-by: ibex-harness-benchmark[bot] <4215784+ibex-harness-benchmark[bot]@users.noreply.github.com>
Signed-off-by: ibex-harness-benchmark[bot] <4215784+ibex-harness-benchmark[bot]@users.noreply.github.com>
Signed-off-by: ibex-harness-benchmark[bot] <4215784+ibex-harness-benchmark[bot]@users.noreply.github.com>
Signed-off-by: ibex-harness-benchmark[bot] <4215784+ibex-harness-benchmark[bot]@users.noreply.github.com>
Signed-off-by: ibex-harness-benchmark[bot] <4215784+ibex-harness-benchmark[bot]@users.noreply.github.com>
Up to standards ✅🟢 Issues
|
✅ All benchmarks passedTL;DR: All collected suites passed. Proxy P99 vs baseline: 📉 -93.9%. Suite matrix
Deep Dive: ProxyLoad test (k6)
Synthetic overhead (Go)
k6 P99 is the SLA. Stage / ns/op values are synthetic Go microbenchmarks, not live traces. Environment & meta
Generated by IBEX Benchmark Bot
|
Signed-off-by: ibex-harness-benchmark[bot] <4215784+ibex-harness-benchmark[bot]@users.noreply.github.com>
Signed-off-by: ibex-harness-benchmark[bot] <4215784+ibex-harness-benchmark[bot]@users.noreply.github.com>
Signed-off-by: ibex-harness-benchmark[bot] <4215784+ibex-harness-benchmark[bot]@users.noreply.github.com>
|
|
There was a problem hiding this comment.
No application code in the PR — skipped Code Health checks.
See analysis details in CodeScene
Quality Gate Profile: Pay Down Tech Debt
Install CodeScene MCP: safeguard and uplift AI-generated code. Catch issues early with our IDE extension and CLI tool.



What and Why
Automated publish of public benchmark history for ibex-harness. Suites in this PR: Proxy, Memory HNSW, Ranking quality, Write pipeline.
Keeps
/benchmarksand/benchmarks/memoryhistory current without contributor-authored commits.Tracking issue
N/A (GitHub App data publish)
How
Proxy suite
Status: ✅ PASS
1.31 ms137.69 µs/opweb/public/benchmarks/benchmark-data.json,web/public/benchmarks/badge.svgk6 p99 is the authoritative SLA metric. Stage breakdown and ns/op values are synthetic Go microbenchmarks.
Memory HNSW suite
1.0000web/public/benchmarks/hnsw-benchmark-data.jsonCorpus cells
Production-like knobs expected (
ef_search=40,min_similarity≈0.70,iterative_scan=off, bulk index build).Ranking-quality suite
1.00001.0000web/public/benchmarks/ranking-quality-benchmark-data.jsonWrite-pipeline suite
33.57 msweb/public/benchmarks/write-pipeline-benchmark-data.jsonSLA: p95 <= 200 ms.
Testing
/benchmarksand/or/benchmarks/memorypreview updatedPerformance
No runtime change — history JSON / badge only. Review suite tables above for SLA deltas.
Security
No secrets; App-signed commits; artifacts re-verified via Actions API before publish.
Migrations / Ops
None.
Docs
Public benchmark history only (
web/public/benchmarks/*). Suite JSON files stay separate.Checklist (Definition of Done)
benchmark-data,hnsw-benchmark-data,ranking-quality-benchmark-data,write-pipeline-benchmark-data)chore/bench-data-publish