✨ Add bsdtar vs pna performance benchmark script

tests/bench/README.md

@@ -0,0 +1,109 @@
+# Performance Benchmarks
+
+Benchmark script to compare processing speed between bsdtar and pna.
+
+## Prerequisites
+
+- [hyperfine](https://github.com/sharkdp/hyperfine) (benchmark runner)
+- `bsdtar` (pre-installed on macOS; `libarchive-tools` package on Linux)
+- `pna` (release-built binary)
+
+```bash
+# macOS
+brew install hyperfine
+
+# Debian/Ubuntu
+sudo apt install hyperfine libarchive-tools
+```
+
+## Quick Start
+
+```bash
+# 1. Build release binary
+cargo build --release -p portable-network-archive
+
+# 2. Run benchmark (small: ~10MB, 200 files)
+./tests/bench/bench_bsdtar_vs_pna.sh --pna target/release/pna
+```
+
+Results are saved to `bench-results-YYYYMMDD-HHMMSS/`.
+
+## Options
+
+| Flag | Default | Description |
+|------|---------|-------------|
+| `--size` | `small` | Dataset size: `small` (10MB), `medium` (100MB), `large` (500MB) |
+| `--pna` | `pna` | Path to pna binary |
+| `--bsdtar` | `bsdtar` | Path to bsdtar binary |
+| `--output-dir` | auto-generated | Output directory for results |
+| `--warmup` | `3` | Number of warmup runs |
+| `--runs` | `5` | Number of measurement runs |
+
+## Examples
+
+```bash
+# Thorough measurement (medium dataset, 10 runs)
+./tests/bench/bench_bsdtar_vs_pna.sh \
+  --pna target/release/pna \
+  --size medium \
+  --runs 10
+
+# Custom output directory
+./tests/bench/bench_bsdtar_vs_pna.sh \
+  --pna target/release/pna \
+  --output-dir my-results
+
+# Minimal run for CI
+./tests/bench/bench_bsdtar_vs_pna.sh \
+  --pna target/release/pna \
+  --size small \
+  --warmup 1 \
+  --runs 2
+```
+
+## What It Measures
+
+### Operations
+
+- **Create**: Archive creation (`bsdtar -cf` vs `pna create -f`)
+- **Extract**: Archive extraction (`bsdtar -xf` vs `pna extract -f`)
+
+### Compression Algorithms
+
+| Algorithm | bsdtar flag | pna flag |
+|-----------|------------|----------|
+| Store (none) | (default) | `--store` |
+| Deflate/gzip | `-z` | `--deflate` |
+| XZ | `-J` | `--xz` |
+| Zstd | `--zstd` | `--zstd` |
+
+On environments where bsdtar lacks zstd support, zstd benchmarks run with pna only.
+
+### Test Data
+
+Deterministically generated files in three categories:
+
+- **Text files (40%)**: Repeated patterns (highly compressible)
+- **Mixed data (30%)**: Pseudo-random data (moderately compressible)
+- **Zero-filled files (30%)**: `/dev/zero` (extremely compressible)
+
+## Output
+
+```
+bench-results-YYYYMMDD-HHMMSS/
+  RESULTS.md              # Markdown summary (environment + all results + size comparison)
+  results/
+    create_store.json     # hyperfine JSON output per benchmark
+    create_store.md       # hyperfine Markdown table
+    create_deflate.json
+    create_deflate.md
+    ...
+```
+
+`RESULTS.md` can be pasted directly into GitHub Issues or PRs.
+
+## Tips
+
+- For accurate results, run with `--runs 10` or more on a quiet system
+- `--size large` takes longer to generate but reveals I/O pattern differences more clearly
+- Use JSON output with `jq` to post-process and compare results