https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69847
Bug ID: 69847
Summary: Spec 2006 403.gcc slows down with -mlra vs. reload on
PowerPC
Product: gcc
Version: 6.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: rtl-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: meissner at gcc dot gnu.org
Target Milestone: ---
I was doing spec 2006 runs on a power8 powerpc little endian box for subversion
id 233194, and I noticed that the 403.gcc benchmark slowed down by about 4.6%
when I used the LRA register allocator, instead the reload register allocator.
Using operf, the first two functions that had significant performance
differences were:
bitmap_element_allocate:
Took 4.08% of the total time with reload, 7.60% with LRA.
bitmap_operation:
Took 5.92% of the total time with reload, 6.62% with LRA.
I will add the various attachments in a bit.
