This message was deleted.

First query:

Just curious, what's the output distribution like for \"colB\" == '123' vs \"colA\" == '123' how many

true

vs

false

for each one?

colA=123 -> 931,795,755 colB=456 -> 2,476,007,551 colB=123 -> 106,878,467 I don’t think the distribution (number of true/false) of these predicates matter though. I tried changing the predicate for colA (in first query) to a value that matches 11,504 true and a value that matches 559,648,183 true. Both completed in around ~10sec

yeah. Agreed. must be something else

Any idea why doesn’t the second query (the faster one) not have the CaseSearchedFunc in the flamegraph?

hmm. don't know.