Vizb is a CLI tool for visualizing Go benchmark results as interactive HTML charts. It helps you compare the performance of different implementations across various workloads.
- Interactive Charts: Generate beautiful, interactive HTML charts from Go benchmark results
- Multiple Metrics: Compare execution time, memory usage, and allocation counts
- Customizable Units: Display metrics in your preferred units (ns/us/ms/s for time, b/B/KB/MB/GB for memory)
- Allocation Units: Customize allocation count representation (K/M/B/T)
- Multiple Output Formats: Generate HTML charts or JSON data with the
--formatflag - Responsive Design: Charts work well on any device or screen size
- Export Capability: Save charts as PNG images directly from the browser
- Simple CLI Interface: Easy-to-use command line interface with helpful flags
- Piped Input Support: Process benchmark data directly from stdin
vizb-varmq-bench.mp4
vizb-varmq-bench-output.mp4
go install github.com/goptics/vizb-
Run your Go benchmarks with JSON output:
go test -bench . -run=^$ -json > bench.json
Important: The input must be a JSON file generated by the
go test -bench -jsoncommand. The tool validates the JSON format and structure before processing. -
Generate a chart from the benchmark results:
vizb bench.json -o output.html
-
Open the generated HTML file in your browser to view the interactive charts.
You can also pipe benchmark results directly to the tool:
go test -bench . -run=^$ -json | vizb -o output.htmlUsage:
vizb [target] [flags]
Arguments:
target Path to benchmark JSON file (optional if using piped input)
Flags:
-h, --help Help for vizb
-n, --name string Name of the chart (default "Benchmarks")
-d, --description string Description of the benchmark
-o, --output string Output HTML file name
-f, --format string Output format (html, json) (default "html")
-s, --separator string Separator for grouping benchmark names (default "/")
-m, --mem-unit string Memory unit available: b, B, KB, MB, GB (default "B")
-t, --time-unit string Time unit available: ns, us, ms, s (default "ns")
-a, --alloc-unit string Allocation unit available: K, M, B, T (default: as-is)
-v, --version Show version informationvizb bench.jsongo test -bench . -run=^$ -json | vizbvizb bench.json -o custom_chart.html# Using long flag name
vizb bench.json --format json -o benchmarks.json
# Using short flag name
vizb bench.json -f json -o benchmark_results.jsonvizb bench.json -n "String Comparison Benchmarks" -d "Comparing different string manipulation algorithms"vizb bench.json -t ms -m MBvizb bench.json -s "_"vizb --versionVizb organizes your benchmark results into groups based on a separator character. By default, this separator is /.
For example, if your benchmark is named BenchmarkTest/workload/subject, Vizb will group it as:
- BenchName:
Test - Workload:
workload - Subject:
subject
You can change the separator using the -s or --separator flag:
vizb bench.json -s "_"With this setting, a benchmark named BenchmarkTest_workload_subject would be grouped the same way.
Vizb always uses:
- The last part after the separator as the
subject(n - 1) - The second-to-last part as the
workload(n - 2) (only ifn > 2) - The first part as the
Testis considered as the name of the benchmark (n - 3)
here n is the number of word parts after the separation.
Important
If n < 3 means BenchmarkTest_workload, then the test is considered as the name of the benchmark and the workload is considered as the name of the subject.
In short, workload is only shows when the n > 2.
Vizb generates three types of charts:
- Execution Time: Shows the execution time per operation for each subject across different workloads
- Memory Usage: Shows the memory usage per operation for each subject across different workloads
- Allocations: Shows the number of memory allocations per operation for each subject
This project uses Task for managing development workflows.
# Install Task runner
go install github.com/go-task/task/v3/cmd/task@latest
# Install required development tools
task setup# List all available tasks
task
# Generate templates
task generate
# Build the binary
task build
# Run tests
task testContributions are welcome! Feel free to open issues or submit pull requests.
This project is licensed under the MIT License - see the LICENSE file for details.
While not required by the MIT license, we kindly ask that you preserve the attribution footer in charts generated by Vizb when displaying them publicly.