https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102948
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Summary|[12 Regression] 60% build |[12 Regression] 60% build
|time regression on gamess |time regression on gamess
|in range |with
|g:31855ba6b16cd138d7484076a |g:31855ba6b16cd138d7484076a
|08cd40d609654b8 |08cd40d609654b8
--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
Yes. Note we're simply vectorizing more and that puts more IL and in
particular more loops (prologue,epilogue,versions) down the pipeline. Nothing
we can do much about apart from making unrelated passes faster.
Unfortunately the extra vectorization doesn't result in a speedup in the end
(it's not to hot code regions).
