docs(11): re-measure the overhead budget, and keep both readings - #43
Draft
codeitlikemiley wants to merge 1 commit into
Draft
docs(11): re-measure the overhead budget, and keep both readings#43codeitlikemiley wants to merge 1 commit into
codeitlikemiley wants to merge 1 commit into
Conversation
Task 9 of the run brief: run the overhead bench, and if the numbers no longer hold, correct the doc rather than the measurement. The budget holds — establishment p99 99.1µs against 3ms, per-frame p99 2.21µs against 1ms, roughly thirty times the headroom. But every figure is about double what docs/11 recorded on 2026-08-20, and that is worth writing down while the margin is large, because that is the only time it is cheap to. **The doubling is not attributed, because it cannot honestly be.** Two things changed. The establishment path gained real work: `ExactCache` went async with a bounded lookup (M11.10), per-route TTL resolution (M11.11), and the classifier's confidence now rides to the audit (M12.5 groundwork). And the host was not quiet — load average ~4 with another application holding ~68% of a core, where the original figures came from an idle machine. Picking either cause would be inventing one. Both are named; a clean re-measurement on an idle host is what would settle it and has not been done. Both dated readings are kept rather than the new one overwriting the old. One number reads as "this is what it is"; two read as "this is how it moves", which is the more useful thing to know about a budget with this much room. Claude-Session: https://claude.ai/code/session_017kFpYDqvz6sKGSkM4YKaRf
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.
Task 9: run the overhead bench; if the numbers no longer hold, correct the doc, not the measurement.
The budget holds with ~30× headroom, and every figure roughly doubled. Both are true, and the second isn't dismissed by the first — a number that moves 2× is worth noticing while the margin is large, because that's the only time it's cheap to.
The doubling is deliberately not attributed
Two things changed between the readings and this measurement cannot separate them:
ExactCachebecame async with a bounded lookup (M11.10), per-route TTL resolution (M11.11), and the classifier's confidence now rides to the audit (M12.5 groundwork, merged an hour ago).Choosing either cause would be inventing one. Both are named, and a clean re-measurement on an idle host is what would settle it. It has not been done.
Both dated readings are kept rather than the new overwriting the old: one number reads as "this is what it is", two read as "this is how it moves".
Doc-only;
fmt --checklocally, CI is the gate.https://claude.ai/code/session_017kFpYDqvz6sKGSkM4YKaRf