[Bug middle-end/63155] [6/7/8/9 Regression] memory hog

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63155

--- Comment #25 from Richard Biener <rguenth at gcc dot gnu.org> ---
Created attachment 44705
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=44705&action=edit
patch

This "simple" one helps.  It builds partition bases similar to the
-ftree-coalesce-vars case.  Correctness needs to be verified still but I think
it errs on the safe side (computing too many dependences at most).  In
particular it avoids computing dependences between things we'll later not try
coalescing anyways.