The vSphere Bitfusion client can run machine learning applications on remote shared GPUs. By using the run command, you can start a single application in vSphere Bitfusion.

The vSphere Bitfusion command to start an application is run with a mandatory argument for the number of the GPUs. To distinguish vSphere Bitfusion arguments from applications, you use a double-hyphen separator or place the application within quotes. You start an application in vSphere Bitfusion by replacing the placeholder values with actual values and running one of the following commands.
  • bitfusion run -n num_gpus other switches -- applications and arguments
  • bitfusion run -n num_gpus other switches "applications and arguments"
By running the run command, you can perform the following three tasks.
  1. Allocate GPUs from the shared pool
  2. Start an application in an environment that can access the GPUs when the application makes CUDA calls
  3. Deallocate the GPUs when the application closes

The run command encapsulates the request_gpus, client, and release_gpus commands. You can use the individual commands to allocate GPUs and run multiple applications on the same GPUs. For more information, see Starting Applications with Reserved GPUs.