Does Taucs have a fortran interface?
tu-...@126.com
Evgenii Rudnyi
> How to use Taucs in a Fortran program? Thx
I am not sure if this is possible. If you are working with Fortran, then
why don't you want to use MUMPS?
余进
platform. It looks a difficult work to compile it.
-----邮件原件-----
发件人: matrixpr...@googlegroups.com
[mailto:matrixpr...@googlegroups.com] 代表 Evgenii Rudnyi
发送时间: 2009年3月21日 4:41
收件人: matrixpr...@googlegroups.com
主题: [matrixprogramming] Re: Does Taucs have a fortran interface?
Evgenii Rudnyi
> Because I use Windows XP, and I don't compile it successfully under this
> platform. It looks a difficult work to compile it.
What a compiler do you use on XP?
余进
-----邮件原件-----
发件人: matrixpr...@googlegroups.com
[mailto:matrixpr...@googlegroups.com] 代表 Evgenii Rudnyi
收件人: matrixpr...@googlegroups.com
主题: Re: 答复: [matrixprogramming] Re: Does Taucs have a fortran interface?
Evgenii Rudnyi
> Visual Studio 2008 + Intel fortran 11
Recently on the MUMPS list there was a discussion on how to use WinMUMPS
to compile MUMPS with these compilers. Alternatively you can use my text
http://matrixprogramming.com/MUMPS/
on how to compile MUMPS with GNUMake.
This should be much easier that to try to call TAUCS from Fortran.
余进
Studio command line, I got a error: makefile<17>, syntax error : expected
':' or '=' separator Stop. I couldn't use Cygwin because of the error:
ifort:command not found
-----邮件原件-----
发件人: matrixpr...@googlegroups.com
[mailto:matrixpr...@googlegroups.com] 代表 Evgenii Rudnyi
收件人: matrixpr...@googlegroups.com
主题: Re: 答复: 答复: [matrixprogramming] Re: Does Taucs have a fortran
interface?
Evgenii Rudnyi
> I followed the steps in your text, but when I used nmake in Microsoft Visual
> Studio command line, I got a error: makefile<17>, syntax error : expected
The formats for GNUmake and NMake are different. It is impossible to use
makefiles directly without editing.
> ':' or '=' separator Stop. I couldn't use Cygwin because of the error:
> ifort:command not found
You have to setup VS for Cygwin. Please look at
http://matrixprogramming.com/Tools/UseVisualStudio.html
Well, you have to extend the variable to work with Intel Fortran.
余进
libdmumps.lib libmumps_common.lib have been made. But I can't built the
example successfully. At first I got an error: you should build the library
../lib/libdmumps.a first. Then I changed .lib to .a
, I got Error 127. Link: Invalid option -- o
-----邮件原件-----
发件人: matrixpr...@googlegroups.com
[mailto:matrixpr...@googlegroups.com] 代表 Evgenii Rudnyi
收件人: matrixpr...@googlegroups.com
主题: Re: 答复: 答复: 答复: [matrixprogramming] Re: Does Taucs have a
fortran interface?
Evgenii Rudnyi
> Tahnk you for you help, I built the lib successfully, the libmpiseq.lib
> libdmumps.lib libmumps_common.lib have been made. But I can't built the
> example successfully. At first I got an error: you should build the library
> ../lib/libdmumps.a first. Then I changed .lib to .a
> , I got Error 127. Link: Invalid option -- o
Have you changed Makefile in examples?
http://matrixprogramming.com/MUMPS/examples/Makefile
It seems that you use the old one. And without changes it does now work
with MS VS.
Alternatively you can use *.lib and compile the example directly with GUI.
Note that you will need METIS lib and BLAS to link MUMPS.
余进
ifort：command line warning #10161: unrecognized source type
'dsimpletest.o'；object file assumed
link: invalid option --o
Is this because I don't link the lib properly?
-----邮件原件-----
发件人: matrixpr...@googlegroups.com
[mailto:matrixpr...@googlegroups.com] 代表 Evgenii Rudnyi
收件人: matrixpr...@googlegroups.com
主题: [matrixprogramming] Re: Does Taucs have a fortran interface?
Evgenii Rudnyi
> I make it again, and now I get this error:
> ifort：command line warning #10161: unrecognized source type
> 'dsimpletest.o'；object file assumed
> link: invalid option --o
The warning can be ignored. But why you have --o, hard to say. Please
have a look at the command. It should actually correspond to this line
$(FL) -o $@ $(OPTL) $@.o $(LIBDMUMPS) $(LIBM) $(LIBBLAS) $(LIBOTHERS)
$(LORDERINGS)
The simplest to find the command, to copy it and then to edit manually.
If you write me the command that you receive, than I can tell you more
exactly what happened.
余进
option --o
But another error happened:
LIBC.lib<crt0dat.obj> :error LNK2005: _exit already defined in
MSVCR90.lib<MSVCR90.dll>
LIBC.lib<fflusht.obj> :error LNK2005: _fflush already defined in
MSVCR90.lib<MSVCR90.dll>
LINK: warning LNK4098: default lib "MSVCRT" conflicts with use of other
libs; use/nodefaultlib:library
Dsimpletest.exe:fatal error LNK1169: one or more multiply defined symbols
found
It seems that I didn't use the right libmetis.lib. Can you send me a copy
which can be used in Windows.
The libmetis.lib I used were MS windows binaries downloaded from METIS's
homepage.
-----邮件原件-----
发件人: matrixpr...@googlegroups.com
[mailto:matrixpr...@googlegroups.com] 代表 Evgenii Rudnyi
收件人: matrixpr...@googlegroups.com
主题: Re: 答复: [matrixprogramming] Re: Does Taucs have a fortran interface?
Evgenii Rudnyi
> I changed FL from ifort to cl, and the I didn't get the error: link: invalid
> option --o
>
> But another error happened:
> LIBC.lib<crt0dat.obj> :error LNK2005: _exit already defined in
> MSVCR90.lib<MSVCR90.dll>
> LIBC.lib<fflusht.obj> :error LNK2005: _fflush already defined in
> MSVCR90.lib<MSVCR90.dll>
> LINK: warning LNK4098: default lib "MSVCRT" conflicts with use of other
> libs; use/nodefaultlib:library
> Dsimpletest.exe:fatal error LNK1169: one or more multiply defined symbols
> found
This is a conflict between -MD and -MT
> It seems that I didn't use the right libmetis.lib. Can you send me a copy
> which can be used in Windows.
> The libmetis.lib I used were MS windows binaries downloaded from METIS's
> homepage.
It is quite simple to compile METIS with MS VS - see
http://matrixprogramming.com/METIS/
-MD is already in the makefile, so it should work.
余进
I have successfully made the metis. Then I type cd to examples dir and type
make, error occurs:
Link: invalid option -- o
Then I chang the FL = ifort to FL = cl, and make, another error occurs:
Link: fatal error LNK 1561: entry point must be defined
Can you take a look at my whole file "MUMPS_4.8.4". If possible, I will
upload it.
-----邮件原件-----
发件人: matrixpr...@googlegroups.com
[mailto:matrixpr...@googlegroups.com] 代表 Evgenii Rudnyi
收件人: matrixpr...@googlegroups.com
主题: Re: 答复: 答复: [matrixprogramming] Re: Does Taucs have a fortran
interface?
余进
Where is the differences between -ML and -MD? Is -MD faster than -ML?
-----邮件原件-----
发件人: matrixpr...@googlegroups.com
[mailto:matrixpr...@googlegroups.com] 代表 Evgenii Rudnyi
收件人: matrixpr...@googlegroups.com
主题: Re: 答复: 答复: [matrixprogramming] Re: Does Taucs have a fortran
interface?
Evgenii Rudnyi
> My problem is solved. I used the default -ML, and then it works.
> Where is the differences between -ML and -MD? Is -MD faster than -ML?
There are three different run-time libraries by MS VS. Well, I think
that now there are only two: it seems that -ML is not supported anymore.
There are problems when one compiles different pieces with different
run-times. Then there is a conflict.
It has nothing to do with the speed but rather with the way how system
libraries are called from the application. You can find more info at
Microsoft Technet.