mako info
The info command provides a detailed overview of the current environment, installed components, GPU devices, Mako configuration paths, available workloads, and setting schema. It is useful for debugging, environment validation, and verifying that all dependencies and configurations are correctly set up before running benchmarks or tuning.
Command Syntax
Optional flags
These flags allow you to narrow the output to specific sections. If no flags are specified, all sections will be shown.
--env- Show detected environment information (OS, Python, packages, visible GPUs, engine availability).--config- Show resolved paths used by Mako for configs, cache, and working directories.--schema- Show the full configuration schema.--tuning- Show the default tuning configuration Mako.--objective- Show the list of available optimization objectives used during tuning. To use one of these for tuning or launching a server, pass --objectivename.
Notes
- You can combine multiple flags to narrow the output.
- If no flags are specified, Mako will show the complete report.
Example output includes:
- System environment info (OS version, Python, ROCm drivers, etc.).
- Installed Python packages relevant to Mako.
- Available GPU devices.
- Mako's home, data, and config directories.
- Available workloads and dataset links (e.g.
sharegpt,rand_10k,rand10_2048). - Missing engines (e.g. if
vllmorsglangare not installed). - Failure to detect tuning settings (if engines are missing or misconfigured).
Typical use cases:
- Checking that your system has ROCm, PyTorch, and other requirements installed.
- Verifying that GPUs are visible and correctly recognized by Mako.
- Listing built-in benchmark workloads and datasets available to your setup.
- Diagnosing issues if benchmark or tuning commands fail.