Documentation for Axxon One 1.0.

Previous page Configuring the acceleration of GPU-based neuroanalytics  Automatic Rules Next page

After the Server is restarted, it may take several minutes for the neural network analytics algorithms to launch on NVIDIA GPU. This time is needed to optimize neural network modules for a specific type of GPU used.

You can use the caching feature so that this operation is performed only once. Caching saves the optimization result to your hard drive and uses it on the next launches of neural network analytics.

To enable caching, create the GPU_CACHE_DIR system variable and in its value specify the existing folder where the optimization result will be stored (see Appendix 10. Creating system variable).

The approximate cache size will depend on the number of neural networks used and their type, the minimum cache size is 70 MB.

Attention!

This feature works on a trial basis.


  • No labels