IBM PowerAI Vision 1.1.2 Release Notes®

Requirements

You must have POWER8® S822LC (8335-GTB) or POWER9™ AC922 with at least one NVIDIA NVLink capable GPU to run IBM® PowerAI Vision. For more information about specific hardware and software requirements, see the Planning for PowerAI Vision topic.

Installing

You can install PowerAI Vision stand-alone or PowerAI Vision with IBM Cloud Private. For more information, see the Installing PowerAI Vision topic.

Limitations

The following are the limitations for IBM PowerAI Vision 1.1.2:
  • If you import a .zip file into an existing data set, the .zip file cannot contain a directory structure.
  • PowerAI Vision uses an entire GPU when you are training a dataset, and when a REST API endpoint is deployed (even if the endpoint is idle). The number of active GPU tasks (model training and deployment) that you can run, at the same time, depends on the number of GPUs on your Power® System server. For example, if your Power Systems™ server has 4 GPUs, you can have a maximum of four training and inference jobs running simultaneously. You must verify that there are enough available GPUs on the system for the desired workload. To check the status of the GPUs, run the nvidia-smi -l command.
  • You cannot install PowerAI Vision stand-alone on the same system that already has IBM Data Science Experience (DSX), IBM Cloud Private, or any other Kubernetes based applications installed.
  • You must uninstall the technology preview version of PowerAI Vision before you can install PowerAI Vision 1.1.2. For more information, see the Uninstalling PowerAI Vision topic.