workerpoh: log effective gpu-chunk / search-timeout-ms in the startup searcher line - #16
Conversation
The startup searcher line (workerpoh: searcher=... mode=... backend=... hybrid_sign=...) is the first thing a miner reads when tuning GPU parameters, but it does not show which -gpu-chunk / -search-timeout-ms are actually in effect. When tuning comes from env defaults or rig profiles instead of explicit flags, a misconfigured value is invisible: the line looks identical whether tuning applied or not, and per-launch overhead at small chunks silently caps hashrate. Append chunk=<N> search_timeout_ms=<N> to the startup line so the effective values are visible immediately, in both cpu and gpu modes. One Fprintf change; behavior otherwise unchanged. Verified: go vet, full package tests, build, and smoke runs (cpu mode) print the new fields; combined with the GPU_CHUNK env-default fix the line reflects env-provided values.
|
Navigate logical layers of code changes, visualize relationships, and explore their blast radius. No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Repository: jokeez/hackme/.coderabbit.yaml Review profile: ASSERTIVE Plan: Advanced Run ID: 📒 Files selected for processing (1)
Included review availability: Your plan provides up to 1 included review per hour; 0 remain after this review. 📝 WalkthroughWalkthroughThe startup searcher log now reports the configured GPU chunk size and per-chunk search timeout in milliseconds. ChangesSearcher startup logging
Priority: ⬇️ Low Estimated code review effort: 1 (Trivial) | ~3 minutes Change: Feature Suggested reviewers: Merge Risk: ⚪ Minimal · up to The startup line makes configured search settings visible; no merge-blocking issue is indicated. Architecture SummaryArchitecture risk: 🔵 Low · up to The change affects 1 system. Changed systems: Architecture concerns Review detailsSystems and components
Before / after behavior
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
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. Comment |
PR Summary by QodoLog effective GPU tuning values at worker startup
AI Description
High-Level Assessment
Files changed (1)
|
Code Review by Qodo🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0)
Great, no issues found!Qodo reviewed your code and found no material issues that require reviewTip of the day💡 Did you know, you can enable the Remediation agent and Qodo fixes findings in a dedicated fix PR |
|
Merged — thanks @bobbyning. Nice follow-up to #14: the startup line now shows the chunk / search_timeout_ms that are actually in effect (including env defaults and flag precedence), so silent misconfig is harder to miss when tuning the 4090 path. CI was green; squash is on main. Pull with rc17.2 / main tip when you upgrade the fleet. |
Problem
The startup line is the first thing a miner reads when tuning GPU parameters:
It does not show which
-gpu-chunk/-search-timeout-msvalues are actually in effect. Since #14 the defaults can come fromGPU_CHUNK/SEARCH_TIMEOUT_MSenvs, but a misconfigured or ignored value is still invisible: the line looks identical whether tuning applied or not, and per-launch overhead at small chunks can silently cap hashrate (on older GPUs the majority of batch time can be launch overhead at the1<<22default).Change
Append
chunk=<N> search_timeout_ms=<N>to the startup line, in both cpu and gpu modes:One
Fprintfchange; behavior otherwise unchanged. Consumers of the line are humans plus archived example logs (docs/archive/vast/*), which append-only fields do not affect; no test or script parses it.Testing
go vet ./cmd/workerpoh/, full package tests, and build — clean (on top of merged workerpoh: honor GPU_CHUNK / SEARCH_TIMEOUT_MS envs for node-spawned workers #14).chunk=4194304 search_timeout_ms=2500GPU_CHUNK=999424 SEARCH_TIMEOUT_MS=777→chunk=999424 search_timeout_ms=777GPU_CHUNK=999424+ explicit-gpu-chunk 3145728→chunk=3145728(flag precedence visible)max(chunk, 1024)) is applied at use time, so a sub-1024 configured chunk prints as configured. In cpu modechunkis configured-but-unused, consistent withbackend=being informational there.Summary by CodeRabbit