Skip to content

docs: Embedded & no_std page + refresh benchmarks to measured data - #25

Merged
ddsha441981 merged 4 commits into
mainfrom
docs/embedded-no-std
Sep 9, 2026
Merged

ddsha441981 merged 4 commits into
mainfrom
docs/embedded-no-std

Conversation

@ddsha441981

@ddsha441981 ddsha441981 commented Sep 9, 2026 •

Copy link
Copy Markdown
Owner

Summary

All numbers in this PR come from runs in this repo's examples (15-trial statistical runs, RSS-based memory measurements). Every claim was re-verified against a fresh full run of the example suite before opening.

Test plan

  • CI green (fmt, clippy, tests, embedded target matrix, loom, miri as configured)
  • mdbook builds locally (mdbook build docs)

- New page: embedded-no-std.md covering target matrix, critical-section,
  QEMU testing, memory footprint (3KB for 64 slots), and PulseMap vs lru
- features.md: add critical-section feature, fix version refs to 0.6,
  cross-link to embedded page
- SUMMARY.md: add Embedded & no_std under Advanced section
- Memory per entry: 25.6B → 34.2B at 1M (reflects v0.6.2 SlotTTL change)
- Memory comparison: 29% less than QuickCache → ~4% (honest number)
- Write p99: 911ns → 853ns (PulseMap improved)
- Head-to-head multipliers recalculated with fresh data
- Scenario B: PulseMap now wins bulk inserts (6.55M vs 5.88M ops/s)
- Scenario C: QuickCache now wins contention (665ns vs 1.301µs)
- benchmarks.md: memory table updated to real measured numbers
- faq.md: memory formula updated to measured 34.2B/entry
Resolve conflicts in CHANGELOG.md and README.md:
- CHANGELOG: keep both AccessBuffer drain + Embedded footprint evidence sections
- README: keep updated Known Limitations with embedded caveats
@ddsha441981
ddsha441981 merged commit 3c3cb88 into main Sep 9, 2026
23 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant