Add benchmarkoor-dashboard: ethrex EL benchmark dashboard#9
Open
edg-l wants to merge 16 commits into
Open
Conversation
…op, scaling; file time_lost counts only deficits
…story, time map); expose in api/agent.md/leaderboard/trends
…threx phase ingestion + fkv
- map ethrex runs to bal-devnet-7 commit (gh history, by time); show in api/agent.md/leaderboard/trends
- lazy /run/{id} block-logs viewer + /api/runs/{id}/block_logs (live, no storage)
- stream benchmarkoor.log per ethrex run: per-test exec/merkle/store phase split + fkv catch-up summary; raw logs discarded
- targets now carry phase bottleneck (exec/merkle/store) + merkle overlap and fkv state in /agent.md + /api/{targets,fkv}
…it (stacked), API fallback otherwise
…ra/hover, calmer section headers, lighter muted, readable prose width, drop dead badge css
- load Inter globally (UI + Plotly chart font), inline svg favicon - desaturate client palette; leaderboard bars = ethrex pink vs muted slate (others) - /run: one clean bar per operation (worst block), fixes segmented-bar collision when ops repeat - verified all pages via headless browser at 2K
- headroom (Mgas/s if ethrex matched best per test) + deficit portfolio by phase/resource
- staleness: store newest-run ts, header colors by snapshot age
- gas-scaling page (/op): Mgas/s vs gas per op, ethrex vs others
- merkle parallelism page (/merkle): serial-merkle ranking from overlap data
- per-commit op regression: phase_history table appended each sync
- failing-tests surfacing from run aggregate
- regression detection /api/regressions + /api/freshness live check
- unauthenticated GitHub REST fallback for commit fetch (incremental cache) when gh absent
- JSON: /api/{headroom,merkle,failures,regressions,freshness}; nav + overview surfaces
…o vs best per gas), drop op dropdown; keep suite picker, op drill-down via ?op=
…'t overflow narrow cards
…ult + full-range toggle
…on heavy tables at limit>=500)
…dexed sort + offset 500s on large run sets)
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
1 participant
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.
New subfolder tool: a FastAPI + Jinja + HTMX + Plotly dashboard over the ethPandaOps Benchmarkoor API, focused on the ethrex EL client.
Sync (
app/sync.py) pulls all suites and results into a local SQLite snapshot. The API intermittently returns 5xx under load, soclient.pywraps every request in tenacity retry/exponential backoff. Only the newest suite per name is indexed (drops stale regenerated duplicates). Run selection picks the latest full run per (suite, instance) to guard against partial/truncated runs skewing aggregates.Pages: Overview, Leaderboard (ranked by gas-weighted aggregate Mgas/s = Σgas/Σtime; median/mean/wins/gas-won as secondary columns), Coverage gaps, Compare (per-test ethrex-vs-clients matrix), Trends, Test detail.
Agent API:
/agent.md(alias/llm.md) is a self-contained Markdown brief for LLM agents. JSON endpoints/api/targets,/api/targets/by_file,/api/leaderboard,/api/coverage,/api/suites. Optimization targets are ranked bytime_lost_ms(ethrex time minus fastest competitor time, summed per file/opcode), which weights by recoverable wall-clock rather than Mgas/s ratio.API key is read from a gitignored
.env(see.env.example). Run:cp .env.example .env, setBENCHMARKOOR_API_KEY,uv sync,uv run python -m app.sync,uv run uvicorn app.main:app.