https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127203
Bug ID: 127203
Summary: [17 Regression] ~30% slowdown of a testcase from
pr26854 since r17-3264-g6cda72b8bd646c
Product: gcc
Version: 17.0
Status: UNCONFIRMED
Keywords: compile-time-hog
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: pheeck at gcc dot gnu.org
CC: aldyh at gcc dot gnu.org
Target Milestone: ---
Host: x86_64-pc-linux-gnu
Target: x86_64-pc-linux-gnu
Created attachment 65491
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=65491&action=edit
the testcase
Since r17-3264-g6cda72b8bd646c, I see ~30% compilation time slowdown on the
attached testcase (originally a testcase for pr26854).
[fkastl@ryzen4 playground]$ binaries/release-6cda72b8bd646c/bin/gcc pr26854-2.c
-O1 -g -ftime-report -c
...
TOTAL : 8.84 314M
[fkastl@ryzen4 playground]$ binaries/release-before-6cda72b8bd646c/bin/gcc
pr26854-2.c -O1 -g -ftime-report -c
...
TOTAL : 6.65 295M
Interesting parts of the time report are:
Time variable wall before after; mem before after; wall difference
phase opt and generate 5.89 8.01 233M 252M 2.12
callgraph functions expansion 5.15 7.09 141M 150M 1.94
callgraph ipa passes 0.65 0.82 27M 36M 0.17
ipa function summary 0.06 0.13 3830k 7065k 0.07
dominator optimization 0.97 2.57 21M 22M 1.6
tree copy headers 0.01 0.2 377k 2338k 0.19
access analysis 0.04 0.14 233k 5454k 0.1
As far as I know, there is no time variable specific to ranger. It seems
probable to me that we are seeing additional ranger computations being
accounted under various time variables.
Aldy, could you look at this? If it seems ok to you, just close the bug as
WON'T FIX. I don't know if this indicates some problem or not but I thought
you might know.
