How to improve SQL lab created virtual dataset charting performance

Question

I have a question regarding SQL lab created virtual dataset performance.

I use SQL lab to generate a table for charting but it often timeouts when I use the same virtual dataset for multiple charts in a dash. I have tried 2 solutions to this problem and have respective questions. The version I am using is Version: 0.999.0dev

Solution 1: I try to use CTAS and CVAS in superset to create a middle layer table for charting to use which loads much faster. But the problem is that it doesn't refresh when new data comes in; would appreciate anyone could advise how to make CTAS and CVAS tables and views created through SQL lab auto-refresh?

Solution 2: I try to set up virtual dataset cache, but the cache is not working on the datasource level, but instead on the chart query level (aka. everytime I change the config in a chart or filter setting in a dash, the virtual dataset basically re-runs). So the cache setting only partially solve the problem.