https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124599
Bug ID: 124599
Summary: [16 Regression] 100% link time slowdown (with LTO) of
459.GemsFDTD since r16-8128-g5f1024922d3f67
Product: gcc
Version: 16.0
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: rguenth at gcc dot gnu.org
Blocks: 26163
Target Milestone: ---
Host: aarch64-gnu-linux, x86_64-pc-linux-gnu
Target: aarch64-gnu-linux, x86_64-pc-linux-gnu
Created attachment 63992
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=63992&action=edit
r16-8127 -ftime-report output
Compile the 459.GemsFDTD SPEC CPU 2006 benchmark with -O2 -flto and measure the
time it takes to link the binary.
Using -flto=64 on my machine I see
On GCC revision r16-8127
real 0m4.374s
user 0m8.594s
sys 0m0.180s
On GCC revision r16-8128
real 0m8.109s
user 0m12.352s
sys 0m0.176s
Commit r16-8128 is:
commit 5f1024922d3f675852196cdd45b47935892c2c96
Author: Richard Biener <[email protected]>
AuthorDate: Fri Mar 13 13:11:56 2026 +0100
Commit: Richard Biener <[email protected]>
CommitDate: Tue Mar 17 13:28:19 2026 +0100
tree-optimization/122380 - avoid PRE/hoisting of incompatible access-path
I've tried doing the linking with -flto=1 -ftime-report. I attach the full
output. This caught my attention:
On GCC revision r16-8127
Time variable wall GGC
alias stmt walking : 0.54 ( 13%) 14M ( 10%)
TOTAL : 4.27 150M
On GCC revision r16-8128
Time variable wall GGC
alias stmt walking : 3.98 ( 50%) 114M ( 44%)
TOTAL : 7.94 258M
Looks like almost all of the additional compilation time is spent in "alias
stmt walking".
Referenced Bugs:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26163
[Bug 26163] [meta-bug] missed optimization in SPEC (2026, 2k17, 2k and 2k6 and
95)
