[Bug tree-optimization/79389] [7 Regression] 30% performance regression in SciMark2 MonteCarlo

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79389

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
So we think the splitting is useful because on one path it exposes that
under_curve is unchanged which might expose a jump threading opportunity.
Not in this case so the possible jump threading opportunity needs to be
further verified by some heuristics (a use in a conditional maybe?).