> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cow.bleu.builders/llms.txt
> Use this file to discover all available pages before exploring further.

# Benchmarking

> Tools and techniques for gas benchmarking and performance tracing of CoW Protocol smart contracts

# Benchmarking

Tools and techniques for measuring and optimizing gas consumption in CoW Protocol smart contracts.

## Benchmark Commands

### Full Suite

Execute the complete benchmark suite across settlement scenarios:

```bash theme={null}
yarn bench
```

### Execution Tracing

Generate detailed execution traces identifying gas consumption patterns:

```bash theme={null}
yarn bench:trace
```

Tracing reveals:

* Call stack hierarchy
* Per-operation gas consumption
* Storage operations
* External calls
* Event emission expenses

### Version Comparison

Compare performance between code versions:

```bash theme={null}
yarn bench:compare [ref]
```

The comparison script performs five steps:

1. Stash current changes
2. Benchmark current code
3. Check out reference commit
4. Benchmark reference code
5. Generate differential report

**Interpreting results:**

* Positive values indicate regressions (increased gas)
* Negative values show improvements (decreased gas)

### Single Scenario

Test individual settlement scenarios:

```bash theme={null}
yarn bench:single
```

## Optimization Strategy

Follow a structured five-phase approach:

1. **Establish baseline** - Measure current gas consumption
2. **Identify hotspots** - Use tracing to find expensive operations
3. **Implement improvements** - Target specific gas-heavy operations
4. **Verify gains** - Compare against baseline measurements
5. **Comprehensive testing** - Ensure correctness is maintained

## Compiler Configuration

The project compiler prioritizes execution costs:

| Setting        | Value     | Rationale                       |
| -------------- | --------- | ------------------------------- |
| Optimizer      | Enabled   | Gas efficiency                  |
| Optimizer Runs | 1,000,000 | Favor execution over deployment |
| EVM Version    | Cancun    | Latest EVM features             |

## Best Practices

* Ensure clean builds without coverage instrumentation before benchmarking
* Use a stable environment (same machine, similar load)
* Run multiple iterations for consistent results
* Compare against known reference points
* Document significant performance changes

<Warning>
  Coverage instrumentation alters contract bytecode and gas costs. Always benchmark against clean, non-instrumented builds.
</Warning>