* Cygwin 32bit: Can't use gcc -mfpmath=sse
@ 2015-09-09 23:30 HK
2015-09-10 11:19 ` Václav Haisman
0 siblings, 1 reply; 6+ messages in thread
From: HK @ 2015-09-09 23:30 UTC (permalink / raw)
To: cygwin
I've just run across this strange behavior on a recent 32bit installation:
vega> cat hello.c
#include <stdio.h>
int main(int argc, char** argv){
printf("hello world\n");
}
vega> gcc -mfpmath=sse hello.c
hello.c:1:0: warning: SSE instruction set disabled, using 387 arithmetics
#include <stdio.h>
^
Why is this? I am pretty sure this used to work in the past. On the 64bit
installtion it works just fine.
For reference:
vega> gcc -v
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/lib/gcc/i686-pc-cygwin/4.9.3/lto-wrapper.exe
Target: i686-pc-cygwin
Configured with:
/cygdrive/i/szsz/tmpp/gcc/gcc-4.9.3-1.i686/src/gcc-4.9.3/configure
--srcdir=/cygdrive/i/szsz/tmpp/gcc/gcc-4.9.3-1.i686/src/gcc-4.9.3
--prefix=/usr --exec-prefix=/usr --localstatedir=/var --sysconfdir=/etc
--docdir=/usr/share/doc/gcc --htmldir=/usr/share/doc/gcc/html -C
--build=i686-pc-cygwin --host=i686-pc-cygwin --target=i686-pc-cygwin
--without-libiconv-prefix --without-libintl-prefix --libexecdir=/usr/lib
--enable-shared --enable-shared-libgcc --enable-static
--enable-version-specific-runtime-libs --enable-bootstrap
--enable-__cxa_atexit --with-dwarf2 --with-arch=i686 --with-tune=generic
--disable-sjlj-exceptions
--enable-languages=ada,c,c++,fortran,java,lto,objc,obj-c++
--enable-graphite --enable-threads=posix --enable-libatomic
--enable-libgomp --disable-libitm --enable-libquadmath
--enable-libquadmath-support --enable-libssp --enable-libada
--enable-libjava --enable-libgcj-sublibs --disable-java-awt
--disable-symvers --with-ecj-jar=/usr/share/java/ecj.jar --with-gnu-ld
--with-gnu-as --with-cloog-include=/usr/include/cloog-isl
--without-libiconv-prefix --without-libintl-prefix --with-system-zlib
--enable-linker-build-id
Thread model: posix
gcc version 4.9.3 (GCC)
vega>
--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Cygwin 32bit: Can't use gcc -mfpmath=sse
2015-09-09 23:30 Cygwin 32bit: Can't use gcc -mfpmath=sse HK
@ 2015-09-10 11:19 ` Václav Haisman
0 siblings, 0 replies; 6+ messages in thread
From: Václav Haisman @ 2015-09-10 11:19 UTC (permalink / raw)
To: cygwin
On 10 September 2015 at 01:30, HK wrote:
> I've just run across this strange behavior on a recent 32bit installation:
>
> vega> cat hello.c
> #include <stdio.h>
> int main(int argc, char** argv){
> printf("hello world\n");
> }
> vega> gcc -mfpmath=sse hello.c
> hello.c:1:0: warning: SSE instruction set disabled, using 387 arithmetics
Does it help to use `-march=native`? My hunch is that this is because
the default CPU type is set to such that does not have SSE.
> #include <stdio.h>
> ^
>
> Why is this? I am pretty sure this used to work in the past. On the 64bit
> installtion it works just fine.
>
> For reference:
>
> vega> gcc -v
> Using built-in specs.
> COLLECT_GCC=gcc
> COLLECT_LTO_WRAPPER=/usr/lib/gcc/i686-pc-cygwin/4.9.3/lto-wrapper.exe
> Target: i686-pc-cygwin
> Configured with:
> /cygdrive/i/szsz/tmpp/gcc/gcc-4.9.3-1.i686/src/gcc-4.9.3/configure
> --srcdir=/cygdrive/i/szsz/tmpp/gcc/gcc-4.9.3-1.i686/src/gcc-4.9.3
> --prefix=/usr --exec-prefix=/usr --localstatedir=/var --sysconfdir=/etc
> --docdir=/usr/share/doc/gcc --htmldir=/usr/share/doc/gcc/html -C
> --build=i686-pc-cygwin --host=i686-pc-cygwin --target=i686-pc-cygwin
> --without-libiconv-prefix --without-libintl-prefix --libexecdir=/usr/lib
> --enable-shared --enable-shared-libgcc --enable-static
> --enable-version-specific-runtime-libs --enable-bootstrap
> --enable-__cxa_atexit --with-dwarf2 --with-arch=i686 --with-tune=generic
> --disable-sjlj-exceptions
> --enable-languages=ada,c,c++,fortran,java,lto,objc,obj-c++ --enable-graphite
> --enable-threads=posix --enable-libatomic --enable-libgomp --disable-libitm
> --enable-libquadmath --enable-libquadmath-support --enable-libssp
> --enable-libada --enable-libjava --enable-libgcj-sublibs --disable-java-awt
> --disable-symvers --with-ecj-jar=/usr/share/java/ecj.jar --with-gnu-ld
> --with-gnu-as --with-cloog-include=/usr/include/cloog-isl
> --without-libiconv-prefix --without-libintl-prefix --with-system-zlib
> --enable-linker-build-id
> Thread model: posix
> gcc version 4.9.3 (GCC)
> vega>
--
VH
--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Cygwin 32bit: Can't use gcc -mfpmath=sse
2015-09-11 12:04 ` HK
@ 2015-09-11 13:40 ` Tim Prince
0 siblings, 0 replies; 6+ messages in thread
From: Tim Prince @ 2015-09-11 13:40 UTC (permalink / raw)
To: cygwin
On 9/11/2015 8:03 AM, HK wrote:
> On Fri, 11 Sep 2015 10:16:56 +0200, Evgeny Grin <k2k@yandex.ru> wrote:
>
>> 10.09.2015, 23:52, "HK" <hk1020@t-online.de>:
>>> On Thu, 10 Sep 2015 13:19:04 +0200, V?clav_Haisman wrote:
>>>> On 10 September 2015 at 01:30, HK wrote:
>>>>> hello.c:1:0: warning: SSE instruction set disabled, using 387
>>>>> arithmetics
>>>>
>>>> Does it help to use `-march=native`? My hunch is that this is because
>>>> the default CPU type is set to such that does not have SSE.
>>>
>>> Yep, that did the trick. Thanks for the suggestion. Now, is this a gcc
>>> build
>>> build problem? The 64bit version doesn't need -march=native and that
>>> is on
>>> the same computer.
>>
>> It's not a problem as by default GCC generate code compatible with
>> maximum number of CPU models.
>> If you need to generate an SSE instructions, you have to use at lest
>> -march=pentium3.
>> For x86-64 version, SSE is always enabled as all x86-64 CPUs support
>> SSE and SSE2.
>> See http://gcc.gnu.org/onlinedocs/gcc/x86-Options.html#x86-Options
>
> Thanks. While I don't quite agree with the choice of defaults it makes
> sense.
> So does my 32bit window next to the 64bit window on the same computer
> really
> have a different instruction set? Anyway, case closed.
Intel compilers made SSE2 the default even for 32-bit mode, subsequent
to all CPUs which supported 387 but not SSE3 going out of production.
There's still a lot of interest in 387 mode, however.
You might argue for making SSE3 the default, but it's generally
important nowadays to set an appropriate option for the target platforms.
--
Tim Prince
--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Cygwin 32bit: Can't use gcc -mfpmath=sse
[not found] ` <1999691441959416@web12o.yandex.ru>
@ 2015-09-11 12:04 ` HK
2015-09-11 13:40 ` Tim Prince
0 siblings, 1 reply; 6+ messages in thread
From: HK @ 2015-09-11 12:04 UTC (permalink / raw)
To: cygwin, Evgeny Grin
On Fri, 11 Sep 2015 10:16:56 +0200, Evgeny Grin <k2k@yandex.ru> wrote:
> 10.09.2015, 23:52, "HK" <hk1020@t-online.de>:
>> On Thu, 10 Sep 2015 13:19:04 +0200, V?clav_Haisman wrote:
>>> On 10 September 2015 at 01:30, HK wrote:
>>>> hello.c:1:0: warning: SSE instruction set disabled, using 387
>>>> arithmetics
>>>
>>> Does it help to use `-march=native`? My hunch is that this is because
>>> the default CPU type is set to such that does not have SSE.
>>
>> Yep, that did the trick. Thanks for the suggestion. Now, is this a gcc
>> build
>> build problem? The 64bit version doesn't need -march=native and that is
>> on
>> the same computer.
>
> It's not a problem as by default GCC generate code compatible with
> maximum number of CPU models.
> If you need to generate an SSE instructions, you have to use at lest
> -march=pentium3.
> For x86-64 version, SSE is always enabled as all x86-64 CPUs support SSE
> and SSE2.
> See http://gcc.gnu.org/onlinedocs/gcc/x86-Options.html#x86-Options
Thanks. While I don't quite agree with the choice of defaults it makes
sense.
So does my 32bit window next to the 64bit window on the same computer
really
have a different instruction set? Anyway, case closed.
--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Cygwin 32bit: Can't use gcc -mfpmath=sse
2015-09-10 20:52 HK
@ 2015-09-10 21:20 ` Yaakov Selkowitz
[not found] ` <1999691441959416@web12o.yandex.ru>
1 sibling, 0 replies; 6+ messages in thread
From: Yaakov Selkowitz @ 2015-09-10 21:20 UTC (permalink / raw)
To: cygwin
On Thu, 2015-09-10 at 22:52 +0200, HK wrote:
> On Thu, 10 Sep 2015 13:19:04 +0200, V?clav_Haisman wrote:
> > On 10 September 2015 at 01:30, HK wrote:
> >> I've just run across this strange behavior on a recent 32bit
> >> installation:
> >>
> >> vega> cat hello.c
> >> #include <stdio.h>
> >> int main(int argc, char** argv){
> >> printf("hello world\n");
> >> }
> >> vega> gcc -mfpmath=sse hello.c
> >> hello.c:1:0: warning: SSE instruction set disabled, using 387
> >> arithmetics
> >
> > Does it help to use `-march=native`? My hunch is that this is because
> > the default CPU type is set to such that does not have SSE.
>
> Yep, that did the trick. Thanks for the suggestion. Now, is this a gcc
> build problem? The 64bit version doesn't need -march=native and that is on
> the same computer.
It's not a bug. SSE and SSE2 are part of the core x86_64 instruction
set.
--
Yaakov
--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: Cygwin 32bit: Can't use gcc -mfpmath=sse
@ 2015-09-10 20:52 HK
2015-09-10 21:20 ` Yaakov Selkowitz
[not found] ` <1999691441959416@web12o.yandex.ru>
0 siblings, 2 replies; 6+ messages in thread
From: HK @ 2015-09-10 20:52 UTC (permalink / raw)
To: cygwin
On Thu, 10 Sep 2015 13:19:04 +0200, V?clav_Haisman wrote:
> On 10 September 2015 at 01:30, HK wrote:
>> I've just run across this strange behavior on a recent 32bit
>> installation:
>>
>> vega> cat hello.c
>> #include <stdio.h>
>> int main(int argc, char** argv){
>> printf("hello world\n");
>> }
>> vega> gcc -mfpmath=sse hello.c
>> hello.c:1:0: warning: SSE instruction set disabled, using 387
>> arithmetics
>
> Does it help to use `-march=native`? My hunch is that this is because
> the default CPU type is set to such that does not have SSE.
>
Yep, that did the trick. Thanks for the suggestion. Now, is this a gcc
build
build problem? The 64bit version doesn't need -march=native and that is on
the same computer.
>> #include <stdio.h>
>> ^
>>
>> Why is this? I am pretty sure this used to work in the past. On the
>> 64bit
>> installtion it works just fine.
>>
>> For reference:
>>
>> vega> gcc -v
>> Using built-in specs.
>> COLLECT_GCC=gcc
>> COLLECT_LTO_WRAPPER=/usr/lib/gcc/i686-pc-cygwin/4.9.3/lto-wrapper.exe
>> Target: i686-pc-cygwin
>> Configured with:
>> /cygdrive/i/szsz/tmpp/gcc/gcc-4.9.3-1.i686/src/gcc-4.9.3/configure
>> --srcdir=/cygdrive/i/szsz/tmpp/gcc/gcc-4.9.3-1.i686/src/gcc-4.9.3
>> --prefix=/usr --exec-prefix=/usr --localstatedir=/var --sysconfdir=/etc
>> --docdir=/usr/share/doc/gcc --htmldir=/usr/share/doc/gcc/html -C
>> --build=i686-pc-cygwin --host=i686-pc-cygwin --target=i686-pc-cygwin
>> --without-libiconv-prefix --without-libintl-prefix --libexecdir=/usr/lib
>> --enable-shared --enable-shared-libgcc --enable-static
>> --enable-version-specific-runtime-libs --enable-bootstrap
>> --enable-__cxa_atexit --with-dwarf2 --with-arch=i686 --with-tune=generic
>> --disable-sjlj-exceptions
>> --enable-languages=ada,c,c++,fortran,java,lto,objc,obj-c++
>> --enable-graphite
>> --enable-threads=posix --enable-libatomic --enable-libgomp
>> --disable-libitm
>> --enable-libquadmath --enable-libquadmath-support --enable-libssp
>> --enable-libada --enable-libjava --enable-libgcj-sublibs
>> --disable-java-awt
>> --disable-symvers --with-ecj-jar=/usr/share/java/ecj.jar --with-gnu-ld
>> --with-gnu-as --with-cloog-include=/usr/include/cloog-isl
>> --without-libiconv-prefix --without-libintl-prefix --with-system-zlib
>> --enable-linker-build-id
>> Thread model: posix
>> gcc version 4.9.3 (GCC)
>> vega>
--
Problem reports: http://cygwin.com/problems.html
FAQ: http://cygwin.com/faq/
Documentation: http://cygwin.com/docs.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2015-09-11 13:40 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-09-09 23:30 Cygwin 32bit: Can't use gcc -mfpmath=sse HK
2015-09-10 11:19 ` Václav Haisman
2015-09-10 20:52 HK
2015-09-10 21:20 ` Yaakov Selkowitz
[not found] ` <1999691441959416@web12o.yandex.ru>
2015-09-11 12:04 ` HK
2015-09-11 13:40 ` Tim Prince
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).