Ori Compiler Benchmarks

Run performance benchmarks for the Ori compiler with duration level: $ARGUMENTS

Duration Levels

Level	Samples	Use Case
`short`	~50 samples, 2s warmup	Quick sanity check (~30s)
`medium`	~100 samples (default)	Standard benchmarking (~2min)
`long`	~300 samples, extended	Release validation (~5min)
`memory`	Single iteration	Heap allocation profiling

Benchmark Suites

Throughput Benchmarks

•parser/raw/throughput/* — Raw parser throughput (MiB/s)
•lexer/raw/throughput/* — Raw lexer throughput (MiB/s)

Memory Benchmarks

•memory/lexer/* — Lexer heap allocations
•memory/parser/* — Parser heap allocations
•memory/summary — Memory amplification report

Instructions

•Build release profile first
•Run benchmarks based on duration level
•Report throughput numbers with comparison to baseline
•Highlight any regressions (>5% slower or >20% more memory)

Benchmark Commands

For short duration:

bash

cargo bench -p oric --bench parser -- "raw/throughput" --noplot --warm-up-time 2 --sample-size 50
cargo bench -p oric --bench lexer -- "raw/throughput" --noplot --warm-up-time 2 --sample-size 50

For medium duration:

bash

cargo bench -p oric --bench parser -- "raw/throughput" --noplot
cargo bench -p oric --bench lexer -- "raw/throughput" --noplot

For long duration:

bash

cargo bench -p oric --bench parser -- "raw" --noplot --sample-size 300
cargo bench -p oric --bench lexer -- "raw" --noplot --sample-size 300

For memory profiling:

bash

cargo bench -p oric --bench memory -- "memory/summary" --noplot

Expected Baselines

See baselines.md for current performance targets.

Output Format

For throughput benchmarks, report as:

code

## Benchmark Results ($ARGUMENTS)

### Parser Throughput
| Workload | Throughput | vs Baseline |
|----------|------------|-------------|
| 10 funcs | XX MiB/s   | +X% / -X%   |
| ...      | ...        | ...         |

### Lexer Throughput
| Workload | Throughput | vs Baseline |
|----------|------------|-------------|
| ...      | ...        | ...         |

### Summary
- Parser: XX MiB/s average
- Lexer: XX MiB/s average
- Status: OK / REGRESSION DETECTED

For memory benchmarks, report the summary table and note any significant changes in allocation count or peak memory.