DAI runs slowly with large model - running 5.3 - any advice


We have developed quite a large model in the system. The model has 41 states and 258 actions so is big. Every time you change the model the autosave kicks in and slows the process leading to 10-20 seconds wait time for actions to be completed. There is a large amount of connectivity between the actions and states, so splitting the model is a challenge.
We havent actually run any major tests yet so there are no large log files floating around. We have had a much smaller model and we dont have issues. The server is 128GB RAM, 2.3GHZ 16 core and is a well spec’d machine.
Splitting the model is a fair chunk of work, so we dont want to have to do this, has anyone else seen this issue and how did you deal with it?

Any help much appreciated.

1 Like

I spoke with the dev team leads about this performance issue, and it is a known issue with larger models. The issue is related to rendering the model in the designer window, and also with the auto-saving feature.

Both of these will be improved in the DAI v6.0, due out at the end of this quarter. In the meantime there is unfortunately no easy workaround. Breaking the model up into smaller components was mentioned. I know that’s not what you want to hear, but v6.0 should be available within the next few weeks and that should help significantly.