- BENCHMARKS_AND_REPRODUCIBILITY.md - Full benchmark suite with reproducibility
- PERFORMANCE_TUNING.md - Parameter optimization guide
- RESOURCE_REQUIREMENTS.md - Hardware recommendations
- LOAD_TESTING.md - Stress test scenarios
- PROFILING_GUIDE.md - CPU, memory profiling instructions
- PERFORMANCE_REGRESSION.md - CI gates and performance thresholds
- Phase 1 (Transport): 2,525 chunks/sec
- Phase 2 (Streaming): 160K+ ops/sec
- Phase 3 (Federation): 10K+ gradients/sec per tier
- Phase 4 (Production): Validated at 100-node scale
See ../INDEX.md for complete documentation navigation.