optimization/10196: [3.3 regression] Compile time regression with inlining

rguenth@tat.physik.uni-tuebingen.de rguenth@tat.physik.uni-tuebingen.de
Mon Mar 24 00:36:00 GMT 2003


>Number:         10196
>Category:       optimization
>Synopsis:       [3.3 regression] Compile time regression with inlining
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Mar 23 22:16:00 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     POOMA library
>Release:        g++-3.3 (GCC) 3.3 20030318 (prerelease)
>Organization:
>Environment:
ix86 linux
>Description:
The testcase at

http://www.tat.physik.uni-tuebingen.de/~rguenth/DynamicLayout2.ii.gz

experiences dramatically slowdown in compile-time compared from the 3.2 release. This is fixed with either specifying -fno-default-inline or --param max-inline-insns-single=200.

The problem was discussed on gcc list in this thread:
http://gcc.gnu.org/ml/gcc/2003-03/msg01174.html
>How-To-Repeat:
compile the testcase with optimization that turns inlining on (-O2).
>Fix:
Can be workarounded by limiting inlining.
>Release-Note:
>Audit-Trail:
>Unformatted:



More information about the Gcc-bugs mailing list