http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52511
Bug #: 52511
Summary: gcc.dg/graphite/interchange-8.c times out on
Solaris/SPARC
Classification: Unclassified
Product: gcc
Version: 4.7.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
AssignedTo: unassig...@gcc.gnu.org
ReportedBy: r...@gcc.gnu.org
CC: ebotca...@gcc.gnu.org, gros...@gcc.gnu.org
Host: sparc-sun-solaris2*
Target: sparc-sun-solaris2*
Build: sparc-sun-solaris2*
I often find that the gcc.dg/graphite/interchange-8.c test times out on
Solaris/SPARC:
The compile time picture is quite mixed (all for 32-bit compiler binaries,
32-bit compilation):
* s8.mayon (SPARC Enterprise T5220, 8-core 1.2 GHz UltraSPARC T2 CPU. i.e.
sun4v):
real 2:08.42
user 2:08.06
sys 0.19
* apoc (Sun Fire V890, 1.35 GHz UltraSPARC IV CPU, i.e. sun4u):
real 59.67
user 59.37
sys 0.19
* for comparison: fuego (ThinkPad W510, 1.6 GHz Core i7 CPU):
real 19.79
user 19.73
sys 0.02
-ftime-report output (attached) shows that compile time is completely dominated
by Graphite data dep analysis.
The question is:
* Could the testcase be reduced to remain well under the default timeout
of 5 minutes?
* Is there some codegen problem leading to the factor 2 slowdown from sun4u
to sun4v?
Rainer
