When the performance of a model is slower than expected, use Performance Advisor to help identify and resolve bottlenecks.
Create a baseline to compare measurements against.
Select the checks you want to run.
Run Performance Advisor with the selected checks and see recommended changes.
Make changes to the model. You can:
Have Performance Advisor automatically apply the changes it recommends.
Use check settings to decide which changes to apply automatically.
Generate advice only. Review and apply changes manually.
After applying changes, Performance Advisor performs a final validation of the model to see how performance has improved.
If the performance improves, the selected checks were successful. The performance check is complete.
If the performance is worse than the baseline, Performance Advisor reinstates the previous settings of the model.