### Install Critcmp with Cargo

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Install the critcmp command-line tool using Cargo. This is the primary method for obtaining the tool.

```bash
$ cargo install critcmp
```

--------------------------------

### Compare Critcmp Baselines from JSON Files

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Use exported JSON files as baselines for comparison by providing their file paths.

```bash
critcmp before.json change.json
```

--------------------------------

### List Available Critcmp Baselines

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Print a list of all baselines that critcmp recognizes.

```bash
critcmp --baselines
```

--------------------------------

### List (Row) Display Mode for Benchmarks

Source: https://context7.com/burntsushi/critcmp/llms.txt

When many baselines or long benchmark names cause the default column layout to wrap, switch to a flat list display that dedicates one section per benchmark group. This mode is useful for visualizing detailed results when the standard output becomes unwieldy.

```bash
critcmp before after extra1 extra2 extra3 --list
```

```bash
# Combine with filter and threshold
critcmp before after extra1 --list -f 'serialize' -t 3
```

--------------------------------

### Compare Multiple Critcmp Baselines

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Compare more than two baselines simultaneously by listing them as arguments.

```bash
critcmp before change1 change2
```

--------------------------------

### Group Benchmarks by Regex Capture Groups

Source: https://context7.com/burntsushi/critcmp/llms.txt

Use a regex with capturing groups to compare benchmarks within a single baseline that have different names but share a common logical identity. The values of all capturing groups form the comparison key; everything outside captures becomes the "variant" label.

```bash
# Given benchmarks "optimized/input1", "naive/input1", "optimized/input2", "naive/input2"
# in baseline "benches", group by the input name:
critcmp benches -g '\w+/(input\d+)'
```

```bash
# Group by a shared suffix across two algorithmic variants
critcmp benches -g '(small|large)_input$'
```

--------------------------------

### Basic Two-Baseline Comparison

Source: https://context7.com/burntsushi/critcmp/llms.txt

Compare two saved Criterion baselines side by side. The fastest result in each group is highlighted in green and marked with rank 1.00.

```bash
# Save a baseline before a code change
carpenter bench -- --save-baseline before

# Make code changes, then save another baseline
carpenter bench -- --save-baseline after

# Compare the two baselines
critcmp before after
```

--------------------------------

### Compare More Than Two Baselines

Source: https://context7.com/burntsushi/critcmp/llms.txt

Compare any number of baselines simultaneously by providing them as arguments. This can be combined with filters and thresholds.

```bash
# Compare four baselines at once
critcmp main pr-123 pr-456 nightly

# Combine with a filter and threshold
critcmp main pr-123 pr-456 -f 'regex' -t 2
```

--------------------------------

### Group Critcmp Benchmarks by Regex Capture Groups

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Compare benchmarks within the same baseline that do not share identical names, by defining matching criteria using regex capture groups. All benchmarks with equivalent captured groups are included in a single comparison.

```bash
critcmp benches -g '\w+/(input1)'
```

--------------------------------

### Export a Baseline to a JSON File

Source: https://context7.com/burntsushi/critcmp/llms.txt

Export a baseline from the Criterion target directory into a portable JSON file for archiving or sharing. These files can later be compared directly.

```bash
# Export baselines to JSON files
critcmp --export before > before.json
critcmp --export after  > after.json

# Later, compare the exported files directly (no target/ dir needed)
critcmp before.json after.json

# Mix file paths and named baselines
critcmp before.json after
```

--------------------------------

### Custom Target Directory for Critcmp

Source: https://context7.com/burntsushi/critcmp/llms.txt

By default critcmp walks up from the current directory looking for a `target/` folder. Override this for workspaces or non-standard build layouts using either the explicit flag or the CARGO_TARGET_DIR environment variable.

```bash
# Explicit flag
critcmp before after --target-dir /path/to/workspace/target
```

```bash
# Environment variable (also respected by Cargo itself)
CARGO_TARGET_DIR=/path/to/workspace/target critcmp before after
```

--------------------------------

### Filter Benchmarks by Regex

Source: https://context7.com/burntsushi/critcmp/llms.txt

Limit the comparison to only benchmark names matching a given regular expression. Non-matching benchmarks are silently excluded.

```bash
# Only show benchmarks whose names contain "parse"
critcmp before after -f 'parse'

# Match benchmarks with names like "search/utf8" or "search/ascii"
critcmp before after -f 'search/(utf8|ascii)'
```

--------------------------------

### Basic Critcmp Usage: Compare Two Baselines

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Compare two saved Criterion baselines. Ensure baselines are saved using 'cargo bench -- --save-baseline <name>' before running this command.

```bash
critcmp before change
```

--------------------------------

### Flatten Critcmp Results to a List

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Display comparison results in a flattened list format instead of the default column-oriented view, useful for very large comparisons.

```bash
critcmp before change1 change2 change3 change4 change5 --list
```

--------------------------------

### Export Critcmp Baseline to JSON

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Export a specific baseline to a JSON file for persistent storage. This JSON file can later be used as a baseline input.

```bash
critcmp --export before > before.json
```

```bash
critcmp --export change > change.json
```

--------------------------------

### Criterion Benchmark JSON Structure

Source: https://context7.com/burntsushi/critcmp/llms.txt

The exported JSON format matches the 'BaseBenchmarks' struct, containing benchmark names and their associated performance estimates.

```json
{
  "name": "before",
  "benchmarks": {
    "parse/small_input": {
      "baseline": "before",
      "fullname": "before/parse/small_input",
      "criterion_benchmark_v1": {
        "group_id": "parse",
        "function_id": "small_input",
        "value_str": null,
        "throughput": null,
        "full_id": "parse/small_input",
        "directory_name": "parse/small_input"
      },
      "criterion_estimates_v1": {
        "mean":   { "confidence_interval": { "confidence_level": 0.95, "lower_bound": 140.1, "upper_bound": 144.5 }, "point_estimate": 142.3, "standard_error": 1.1 },
        "median": { "confidence_interval": { "confidence_level": 0.95, "lower_bound": 141.0, "upper_bound": 143.8 }, "point_estimate": 142.0, "standard_error": 0.9 },
        "median_abs_dev": { "confidence_interval": { "confidence_level": 0.95, "lower_bound": 0.5, "upper_bound": 2.1 }, "point_estimate": 1.1, "standard_error": 0.4 },
        "slope": null,
        "std_dev": { "confidence_interval": { "confidence_level": 0.95, "lower_bound": 0.8, "upper_bound": 1.5 }, "point_estimate": 1.1, "standard_error": 0.2 }
      }
    }
  }
}
```

--------------------------------

### Suppress Small Differences with a Threshold

Source: https://context7.com/burntsushi/critcmp/llms.txt

Hide comparisons where the performance difference is below a given percentage. Useful for cutting through noise and focusing on regressions or improvements that matter.

```bash
# Only show benchmarks with more than 5% difference
critcmp before after -t 5

# Only show benchmarks with more than 10% difference
critcmp before after -t 10
```

--------------------------------

### Filter Critcmp Comparisons by Regex

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Limit comparisons to benchmarks matching a specific regex pattern using the -f or --filter flag.

```bash
critcmp before change -f 'foo.*bar'
```

--------------------------------

### Filter Critcmp Comparisons by Percentage Difference

Source: https://github.com/burntsushi/critcmp/blob/master/README.md

Hide comparisons with small differences (e.g., 5% or less) using the -t flag.

```bash
critcmp before change -t 5
```

--------------------------------

### Control ANSI Color Output in Critcmp

Source: https://context7.com/burntsushi/critcmp/llms.txt

Control ANSI color output. Defaults to `auto` (color only when stdout is a TTY). Use `always` to force color even when piping, or `never` for plain text output without escape codes.

```bash
critcmp before after --color always   # force color even when piping
```

```bash
critcmp before after --color never    # plain text, no escape codes
```

```bash
critcmp before after --color auto     # default: color when TTY
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.