[mpich-discuss] Compiler optimization flags in MPICH 3.0 rc1
Yauheni Zelenko
zelenko at cadence.com
Fri Nov 16 11:52:34 CST 2012
Hi, Darius!
But why warnings flags are mixed with optimization ones? I think these options should be independent.
Eugene.
________________________________________
From: discuss-bounces at mpich.org [discuss-bounces at mpich.org] On Behalf Of Darius Buntinas [buntinas at mcs.anl.gov]
Sent: Friday, November 16, 2012 9:41 AM
To: discuss at mpich.org
Cc: mpich-discuss at mcs.anl.gov
Subject: Re: [mpich-discuss] Compiler optimization flags in MPICH 3.0 rc1
The --enable-strict flag adds -O2. Try --enable-strict=noopt.
-d
On Nov 15, 2012, at 7:06 PM, Yauheni Zelenko wrote:
> Hi!
>
> I tried to build optimize and debug version of MPICH 3.0 rc1 and noticed some problems with compiler optimization flags. Looks like -O2 is set implicitly.
>
> For optimize configuration command like contains -O2 and -O3:
>
> ./configure --prefix=${InstallDir} \
> --with-pm=hydra --with-device=ch3:nemesis \
> --disable-chekpointing \
> --enable-strict --enable-fast=O3 \
> --disable-f77 --disable-fc \
> CC=${GCCDir}/bin/gcc CFLAGS="-m32 -fPIC" \
> CPP=${GCCDir}/bin/cpp \
> CXX=${GCCDir}/bin/g++ CXXFLAGS="-m32 -fPIC" \
> CXXCPP=${GCCDir}/bin/cpp
>
> For debug one -O2, but optimization is disabled:
>
> ./configure --prefix=${InstallDir} \
> --with-pm=hydra --with-device=ch3:nemesis \
> --disable-chekpointing \
> --enable-strict --disable-fast -enable-g=debug \
> --disable-f77 --disable-fc \
> CC=${GCCDir}/bin/gcc CFLAGS="-m32 -fPIC" \
> CPP=${GCCDir}/bin/cpp \
> CXX=${GCCDir}/bin/g++ CXXFLAGS="-m32 -fPIC" \
> CXXCPP=${GCCDir}/bin/cpp
>
> Eugene.
> _______________________________________________
> discuss mailing list discuss at mpich.org
> To manage subscription options or unsubscribe:
> https://lists.mpich.org/mailman/listinfo/discuss
_______________________________________________
discuss mailing list discuss at mpich.org
To manage subscription options or unsubscribe:
https://lists.mpich.org/mailman/listinfo/discuss
More information about the discuss
mailing list