[Bug middle-end/64928] [8/9/10/11 Regression] Inordinate cpu time and memory usage in "phase opt and generate" with -ftest-coverage -fprofile-arcs

[rguenth at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64928

--- Comment #38 from Richard Biener <rguenth at gcc dot gnu.org> ---
Created attachment 50354
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=50354&action=edit
SVG of the CFG at LIM

This is a SVG of the CFG as created by dot at the point of the first LIM pass.

The CFG isn't too special and I guess a switch instead of the computed goto
would present us with the same issues.

I suppose putting a hard limit on the number of stores to move and then
ordering candidates based on their importance (execution frequency) is the
way to go.