wifi-densepose/vendor/midstream/benches/QUICK_REFERENCE.md

Benchmark Quick Reference Card

🚀 Run Benchmarks

# All benchmarks
./scripts/run_benchmarks.sh

# Individual crate
cargo bench --bench temporal_bench      # Temporal Compare
cargo bench --bench scheduler_bench     # Nanosecond Scheduler
cargo bench --bench attractor_bench     # Attractor Studio
cargo bench --bench solver_bench        # Neural Solver
cargo bench --bench meta_bench          # Strange Loop
cargo bench --bench quic_bench          # QUIC Multistream

# Specific group
cargo bench --bench temporal_bench dtw
cargo bench --bench scheduler_bench overhead

📊 Performance Targets Cheatsheet

Benchmark	Key Metric	Target
temporal_bench	DTW n=100	<10ms
	LCS n=100	<5ms
	Edit n=100	<3ms
scheduler_bench	Schedule	<100ns
	Execute	<1μs
	Stats	<10μs
attractor_bench	Phase space	<20ms
	Lyapunov	<500ms
	Detection	<100ms
solver_bench	Encode	<10ms
	Verify	<100ms
	Parse	<5ms
meta_bench	Learn	<50ms
	Extract	<20ms
	Integrate	<100ms
quic_bench	Stream	<1ms
	Multiplex	<100μs
	Throughput	>1GB/s

🎯 Common Commands

# Compare branches
./scripts/benchmark_comparison.sh main feature-branch

# Save baseline
cargo bench -- --save-baseline main

# Compare with baseline
cargo bench -- --baseline main

# View HTML reports
open target/criterion/*/report/index.html

# View summary
cat target/criterion/SUMMARY.md

🔍 Profiling

# Flamegraph
cargo flamegraph --bench temporal_bench

# perf
perf record -g cargo bench --bench temporal_bench
perf report

# Valgrind
valgrind --tool=cachegrind target/release/deps/temporal_bench-*

📈 Benchmark Groups

temporal_bench

• dtw_benches - DTW performance
• lcs_benches - LCS algorithms
• edit_benches - Edit distance
• cache_benches - Cache scenarios
• memory_benches - Memory patterns

scheduler_bench

• overhead_benches - Schedule overhead
• latency_benches - Task execution
• queue_benches - Priority queue
• stats_benches - Statistics
• threading_benches - Multi-threading

attractor_bench

• embedding_benches - Phase space
• lyapunov_benches - Lyapunov calc
• detection_benches - Attractor detection
• trajectory_benches - Trajectory analysis
• dimension_benches - Dimension estimation
• chaos_benches - Chaos detection
• pipeline_benches - Complete analysis

solver_bench

• encoding_benches - Formula encoding
• parsing_benches - Formula parsing
• verification_benches - Trace verification
• state_benches - State operations
• neural_benches - Neural verification
• operator_benches - Temporal operators
• pipeline_benches - Complete pipeline

meta_bench

• learning_benches - Meta-learning
• pattern_benches - Pattern extraction
• hierarchy_benches - Multi-level learning
• integration_benches - Cross-crate
• recursive_benches - Self-referential
• pipeline_benches - Complete cycle

quic_bench

• stream_benches - Stream operations
• multiplexing_benches - Multiplexing
• connection_benches - Connection setup
• throughput_benches - Data throughput
• concurrent_benches - Concurrent streams
• error_benches - Error handling

📂 File Locations

/workspaces/midstream/
├── benches/
│   ├── temporal_bench.rs        (~450 lines)
│   ├── scheduler_bench.rs       (~520 lines)
│   ├── attractor_bench.rs       (~480 lines)
│   ├── solver_bench.rs          (~490 lines)
│   ├── meta_bench.rs            (~500 lines)
│   ├── quic_bench.rs            (~420 lines)
│   ├── README.md                (Overview)
│   └── QUICK_REFERENCE.md       (This file)
├── scripts/
│   ├── run_benchmarks.sh        (Run all)
│   └── benchmark_comparison.sh  (Compare)
├── docs/
│   └── BENCHMARK_GUIDE.md       (Full guide)
├── BENCHMARKS_SUMMARY.md        (Summary)
└── Cargo.toml                   (Config)

🎨 Output Interpretation

Terminal Colors

• 🟢 Green: Performance improved
• 🟡 Yellow: Within noise threshold
• 🔴 Red: Performance regressed

Key Metrics

• Mean: Average execution time
• Std Dev: Consistency (lower is better)
• Median: Central tendency
• Throughput: Operations/second

⚡ Best Practices

1. Close unnecessary apps before benchmarking
2. Run multiple times for consistency
3. Check std dev - high values indicate noise
4. Use baselines for regression detection
5. Profile hotspots when optimizing
6. Document changes that affect performance

🔧 Troubleshooting

High Variance

# Increase sample size
cargo bench -- --sample-size 200

# Disable frequency scaling
sudo cpupower frequency-set --governor performance

Slow Benchmarks

# Reduce measurement time
cargo bench -- --measurement-time 5

# Run specific tests only
cargo bench --bench temporal_bench dtw_performance/linear/10

Memory Issues

# Run with more memory
cargo bench --release -- --test-threads=1

# Profile memory usage
valgrind --tool=massif cargo bench

📊 Example Output

dtw_performance/linear/100
                        time:   [8.234 ms 8.567 ms 8.912 ms]
                        thrpt:  [11.22 Kelem/s 11.67 Kelem/s 12.14 Kelem/s]
                 change:
                        time:   [-5.2341% -3.1234% -1.0123%] (p = 0.00 < 0.05)
                        thrpt:  [+1.0226% +3.2345% +5.5234%]
                        Performance has improved.

🎯 Quick Wins

1. First time? Run: ./scripts/run_benchmarks.sh
2. Optimizing? Profile: cargo flamegraph --bench <name>
3. Comparing? Run: ./scripts/benchmark_comparison.sh main feature
4. Debugging? Check: target/criterion/*/report/index.html

---

Total Benchmarks: 6 files | Total Lines: ~2,860 | Groups: 45+ | Scenarios: 150+