[Bug target/59601] New: [4.9 Regression] __attribute__ ((target("arch=corei7"))) won't match Westmere processor
hjl.tools at gmail dot com
gcc-bugzilla@gcc.gnu.org
Thu Dec 26 05:30:00 GMT 2013
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59601
Bug ID: 59601
Summary: [4.9 Regression] __attribute__
((target("arch=corei7"))) won't match Westmere
processor
Product: gcc
Version: 4.9.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: hjl.tools at gmail dot com
CC: ubizjak at gmail dot com
g++.dg/ext/mv1.C fails on Westmere processor at run-time
since -march=westmere wasn't supported befor and now
__attribute__ ((target("arch=corei7"))) is turned into
__attribute__ ((target("arch=nehalem"))) which won't match
Westmere at run-time. We have 2 choices:
1. No change. __attribute__ ((target("arch=corei7"))) won't
match Westmere and function won't be optimized for Westmere.
2. Make PROCESSOR_NEHALEM to match "corei7" for function
versioning. But __attribute__ ((target("arch=nehalem"))) function
may be used on Westmere. I think it is OK since function
versioning doesn't support extra ISAs on Westmere.
More information about the Gcc-bugs
mailing list