Geekbench AI Command Line Tool

The Geekbench AI command-line tool called banff provides the same functionality as the graphical user interface. This tool allows you to use Geekbench AI from a command prompt and automate it with shell scripts or batch files. This is the primary interface for the Linux version of Geekbench AI, and is available as a feature of Geekbench AI Pro for Windows and macOS.

Install the Tool on macOS

  • Click on the "Tools" menu.
  • Click on the "Install Terminal Command" menu item.

A dialog box will appear asking you to authenticate as an administrator; enter an administrator's username and password and click "Create Links". Another dialog box will appear with usage instructions (if the install was successful) or an error message (if the install encountered a problem). To use the tool enter the following command in a Terminal window:

banff

Technical Details: The banff command line tool is found inside the Geekbench AI application bundle. Installing the tool creates a symlink to the tool in /usr/local/bin. If you move Geekbench AI to another location you will need to re-install the tool as the symlink will point to an incorrect location.

Install the Tool on Windows

  • Click on the "Tools" menu.
  • Click on the "Install Command Prompt Tool" menu item.

A dialog box will appear with usage instructions (if the install was successful) or an error message (if the install encountered a problem). To use the tool open a new command prompt window (command prompt windows that are already open won't be able to find the tool) and enter the following command:

banff

Technical Details: The banff command line tool is found inside the Geekbench AI install directory. Installing the tool adds the Geekbench AI install directory to your PATH environment variable.

Using the Tool

You can get help at any time with the --help switch:

banff --help

If you don't specify any command-line switches, Geekbench AI runs the CPU benchmark.

Geekbench AI displays its progress while it's running the workloads, and displays the section scores and overall scores once all the workloads have run. Geekbench AI will prompt you as to whether you want to upload the results to the Geekbench Browser.

If you don't want to be prompted (for example, you're running Geekbench AI as part of a script) you can answer this question ahead of time on the command line with one of the following switches:

  • --upload will automatically upload the result to the Geekbench Browser once the benchmark is complete.
  • --no-upload will not upload the result to the Geekbench Browser once the benchmark is complete.

You can save Geekbench AI results with the following command line (note that .gbai is the recommended file extension for Geekbench AI results):

banff --save result.gbai

You can also export Geekbench AI results to different formats. This is a great way to share your results with people who don't have Geekbench AI installed. Exporting is done with one of the following switches:

  • --export-csv exports the result as a CSV file.
  • --export-html exports the result as an HTML file.
  • --export-json exports the result as a JSON file.
  • --export-text exports the result as a plain-text file.

You can load and display a saved Geekbench AI result with the following command line:

banff --load result.gbai

When loading a result, you can manipulate it like any other result. You can upload the saved Geekbench AI result to the Geekbench Browser by using the following switches:

banff --load result.gbai --upload

You can also export the saved Geekbench AI result in a different format. For example, the following command line with export the saved result as an HTML file:

banff --load result.gbai --export-html

You can view the GPU API available on your system with the following command:

banff --gpu-list

You can then select one GPU compute API to run a benchmark with. For example, we can run an OpenCL benchmark on a system that supports OpenCL with the following command:

banff --gpu OpenCL

If you have multiple GPUs which support the same GPU API, you can differentiate between them with the platform and device ID numbers that appear before the GPU's name after running banff --gpu-list. For example, if the output of banff --gpu-list contains two devices under OpenCL, including 0 1 AMD Radeon VII, you can run the OpenCL benchmark on the Radeon VII with the following command:

banff --gpu OpenCL --gpu-platform-id 0 --gpu-device-id 1

Troubleshooting

macOS

If you receive a -bash: banff: command not found error message when trying to run Geekbench AI, either the Geekbench AI command line tool wasn't installed correctly, or the Geekbench AI application bundle was moved. Please re-install the command-line tool.

If the problem persists, it's possible the banff link exists but points to the wrong location and Geekbench AI is unable to remove the incorrect link. Open the Terminal and run ls -al /usr/local/bin/banff to determine if the link was installed correctly. If the link exists remove the link with sudo rm -rf /usr/local/bin/banff.

If the problem still persists, please contact support.

Windows

If you receive a 'banff.exe' is not recognized as an internal or external command, operable program, or batch file. error message when trying to run Geekbench AI, the Geekbench AI command line tool wasn't installed correctly, or the command prompt window was opened before the command line tools were installed. Please close all open command prompt windows and open a new command prompt window. If this does not resolve the issue, please re-install the command-line tool and open a new command prompt window.

If the problem still persists, please contact support.