Hello @Thu Mai !

Long story short, a sparse engine does not take into account the blank cells, therefore it can perform better and faster compared to other calculation engines (where blanks are filled - densified - with 0 or falses).

Please take look in this article for more details.

Let me know if you have any questions!

Thanks!

Igor

Hi!
While a sparse engine is indeed optimised when you do not need all intersections of your dimensions, cells do still count, and if you really want to optimise your model you should try to compute as little as possible.

You’ll find a guide below on how to add the number of cells in each metric in the block explorer.

If you feel a particular block is especially slow, it might be useful to try to analyse whether you absolutely need all cells. 

Hope that helps too!

oh got it! thanks @Igor P and @francois a lot!!