Metadata-Version: 2.1
Name: backend.ai-accelerator-cuda-mock
Version: 22.9.0b4
Summary: Backend.AI Accelerator Plugin for CUDA (Mockup)
Home-page: https://github.com/lablup/backend.ai
Author: Lablup Inc. and contributors
License: LGPLv3
Project-URL: Documentation, https://docs.backend.ai/
Project-URL: Source, https://github.com/lablup/backend.ai
Platform: UNKNOWN
Classifier: Intended Audience :: Developers
Classifier: Operating System :: MacOS :: MacOS X
Classifier: Operating System :: POSIX :: Linux
Classifier: Programming Language :: Python
Classifier: Programming Language :: Python :: 3
Classifier: Environment :: No Input/Output (Daemon)
Classifier: Topic :: Scientific/Engineering
Classifier: Topic :: Software Development
Classifier: Development Status :: 4 - Beta
Classifier: License :: OSI Approved :: GNU Lesser General Public License v3 or later (LGPLv3+)
Requires-Python: ==3.10.5
Description-Content-Type: text/markdown

# backend.ai-accelerator-cuda-mock

A mockup plugin for CUDA accelerators

This plugin deceives the agent and manager to think as if there are CUDA devices.
The configuration follows `cuda-mock.toml` placed in the same location of `agent.toml`.
Please refer the sample configurations in the `configs/accelerator` directory and copy one of them as a starting point.

The statistics are randomly generated in reasonable ranges, but it may seem like "jumping around" because there is no smoothing mechanism of generated values.
The configurations for fractional/discrete mode, fraction size, and device masks in etcd are exactly same as the original plugin.

The containers are created without any real CUDA device mounts but with `BACKENDAI_MOCK_CUDA_DEVICES` and `BACKENDAI_MOCK_CUDA_DEVICE_COUNT` environment variables.
Since the manager does not know if the reported devices are real or not, you can start any CUDA-only containers (but of course they won't work as expected).


