alee/stegasoo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update plan: mark #6 Smoke Test Benchmarking as done

Browse Source 
🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Aaron D. Lee
2026-01-05 19:15:05 -05:00
parent f69475b406
commit 7a35ac3df7
1 changed files with 10 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
PLAN-4.1.2.md
View File

@@ -169,20 +169,21 @@ Polish and UX improvements after the 4.1.1 stability release.
## 6. Smoke Test Benchmarking ## 6. Smoke Test Benchmarking
**Status:** Planned **Status:** Done
**Problem:** No way to measure encode/decode performance or track regressions. **Problem:** No way to measure encode/decode performance or track regressions.
**Solution:** Add timing to smoke tests using `hyperfine` or `time`. **Solution:** Add timing to smoke tests using `hyperfine` or `time`.
### Features ### Implementation
- Benchmark encode/decode on test images - Added `--benchmark` flag to run encode/decode benchmarks after tests
- Output timing stats (min/max/avg) - Added `--runs=N` flag to customize number of benchmark runs (default: 5)
- Optional `--benchmark` flag for smoke-test.sh - Uses hyperfine if available for precise timing with warmup
- Compare NVMe vs SD card, overclocked vs stock - Falls back to manual timing with bc if hyperfine not installed
- Outputs min/max/avg stats for both encode and decode operations
### Files to Modify ### Files Modified
- `rpi/smoke-test.sh` - `tests/smoke-test.sh`
--- ---
@@ -244,6 +245,6 @@ Polish and UX improvements after the 4.1.1 stability release.
## Notes ## Notes
- Keep 4.1.2 focused - 9 features (5 done) - Keep 4.1.2 focused - 9 features (6 done)
- Don't break DCT compatibility (4.1.1 RS format is stable) - Don't break DCT compatibility (4.1.1 RS format is stable)
- Test on Pi before release - Test on Pi before release