Add Performance Benchmarking Suite

[h-mayr]

Goal

Establish automated performance benchmarks for critical operations to catch performance regressions early.

Current State

• No performance testing exists
• Fitting operations involve C++ ROOT code — performance depends on ROOT version, compiler optimizations
• Matrix operations on 1024×1024 files have unknown performance characteristics
• No baseline measurements to compare against

Proposed Solution

Setup

1. Add pytest-benchmark to dev dependencies in pyproject.toml
2. Create tests/benchmarks/conftest.py with benchmark fixtures
3. Establish baseline JSON storage in tests/benchmarks/baselines/

Core Benchmarks
Create benchmark modules:

# tests/benchmarks/bench_io.py
def test_spectrum_load_small(benchmark):  # 1024 channels
    benchmark(specreader.ReadSpectrum, "test_small.spc")

def test_spectrum_load_medium(benchmark):  # 8192 channels
    benchmark(specreader.ReadSpectrum, "test_medium.spc")

def test_matrix_load(benchmark):  # 1024×1024
    benchmark(matrix.LoadMatrix, "test_matrix.mtx")

# tests/benchmarks/bench_fit.py
def test_single_peak_fit(benchmark):
    benchmark(fit.FitPeakFunc, spectrum)

def test_multi_peak_fit_10(benchmark):
    # spectrum with 10 peaks
    benchmark(fit.FitPeakFunc, spectrum_10peaks)

# tests/benchmarks/bench_matrix.py
def test_matrix_projection(benchmark):
    benchmark(matrix.Project, axis=0)

def test_matrix_cut(benchmark):
    benchmark(matrix.ApplyCut, cut_coords)

CI Integration

• Add GitHub Action that runs benchmarks on main branch pushes
• Compare against baseline, comment on PR if >20% regression
• Store benchmark history for trend analysis

Files to Create

• tests/benchmarks/__init__.py
• tests/benchmarks/conftest.py
• tests/benchmarks/bench_io.py
• tests/benchmarks/bench_fit.py
• tests/benchmarks/bench_matrix.py
• tests/benchmarks/baselines/*.json

Metric

• Benchmarks run in <3 minutes on CI
• Performance report generated for each PR to main
• Alert on >20% regression

Impact

• Catch performance regressions before merge (e.g., a change that makes fitting 2x slower)
• Establish baseline for future optimization work
• Document expected performance characteristics for users