I'm brand new to the Einstein Toolkit and to Cactus, so be gentle with me.
I'm getting some messages similar to
"CST error 1: -> Configuration script for thorn BLAS returned exit code 1 Error message: 'Building BLAS requires a Fortran compiler, but there is none configured: F77='none'. Aborting.
I looked to the sl6.cfg file and both F77 and F90 are set to gfortran, and gfortran is installed
Thanks.
Is there another configuration file somewhere that needs to be updated?
Joe
On Dec 10, 2012, at 2:14 AM, Joe Hellmers wrote:
I'm brand new to the Einstein Toolkit and to Cactus, so be gentle with me.
I'm getting some messages similar to
"CST error 1: -> Configuration script for thorn BLAS returned exit code 1 Error message: 'Building BLAS requires a Fortran compiler, but there is none configured: F77='none'. Aborting.
I looked to the sl6.cfg file and both F77 and F90 are set to gfortran, and gfortran is installed
Thanks.
Is there another configuration file somewhere that needs to be updated?
Hi Joe,
can you post the exact command that you're using to build your code? If you're building with simfactory, you need to give it some information regarding which optionlist you want to use. The most straightforward way to do this is by adding an
--optionlist=sl6.cfg
to your build command. Otherwise you can teach it to automatically recognize your machine by creating a yourmachine.ini file under simfactory/mdb/machines, and setting the default optionlist in there.
Does this help?
Eloisa
After adding the optionlist for sl6.cfg, I'm not getting the same error, but I am getting
sol_elliptic_sin_zec.C:56:31: error: gsl/gsl_sf_bessel.h: No such file or directory make[5]: *** [sol_elliptic_sin_zec.o] Error 1 make[4]: *** [install] Error 2 make[3]: *** [Source/Non_class_members/PDE] Error 2 make[3]: *** Waiting for unfinished jobs.... make[2]: *** [cpp] Error 2
CST error 1: -> Configuration script for thorn LORENE returned exit code 1 (no error message)
and then at the end
------------------------------------------------------ There were 2 errors during execution of the CST These must be corrected before compilation can proceed ------------------------------------------------------
------------------------------------------------------ Warnings were generated during execution of the CST ------------------------------------------------------
CST error 1: -> Configuration script for thorn LORENE returned exit code 1 (no error message)
CST error 2: -> Configuration script for thorn HWLOC returned exit code 1 (no error message)
------------------------------------------------------
make[1]: *** [/home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/config-data/make.thornlist] Error 1 make: *** [sim] Error 2
On Dec 10, 2012, at 12:46 AM, Eloisa Bentivegna wrote:
On Dec 10, 2012, at 2:14 AM, Joe Hellmers wrote:
I'm brand new to the Einstein Toolkit and to Cactus, so be gentle with me.
I'm getting some messages similar to
"CST error 1: -> Configuration script for thorn BLAS returned exit code 1 Error message: 'Building BLAS requires a Fortran compiler, but there is none configured: F77='none'. Aborting.
I looked to the sl6.cfg file and both F77 and F90 are set to gfortran, and gfortran is installed
Thanks.
Is there another configuration file somewhere that needs to be updated?
Hi Joe,
can you post the exact command that you're using to build your code? If you're building with simfactory, you need to give it some information regarding which optionlist you want to use. The most straightforward way to do this is by adding an
--optionlist=sl6.cfg
to your build command. Otherwise you can teach it to automatically recognize your machine by creating a yourmachine.ini file under simfactory/mdb/machines, and setting the default optionlist in there.
Does this help?
Eloisa
On Mon, Dec 10, 2012 at 03:56:08PM -0600, Frank Loeffler wrote:
Try installing the gsl-devel package. This should have been caught by the GSL configuration script though.
Having said that, sl6.cfg specifies a GSL dir (and sets it to /usr). I don't think this is a good idea. If gsl is installed in /usr the configure script should be able to find it (and catch your problem) and then build the included gsl version. Should this definition be removed or is there a reason for it to be there?
Frank
On Mon, Dec 10, 2012 at 4:59 PM, Frank Loeffler knarf@cct.lsu.edu wrote:
On Mon, Dec 10, 2012 at 03:56:08PM -0600, Frank Loeffler wrote:
Try installing the gsl-devel package. This should have been caught by the GSL configuration script though.
Having said that, sl6.cfg specifies a GSL dir (and sets it to /usr). I don't think this is a good idea. If gsl is installed in /usr the configure script should be able to find it (and catch your problem) and then build the included gsl version. Should this definition be removed or is there a reason for it to be there?
You're right. It's probably better to not have it in there in case the user doesn't have the gsl-devel package installed. I have removed this (and some other similarly unnecessary options) from the optionlist. I also fixed a bug in the optionlist which prevented it from working with fortran code.
Joe, can you try the latest sl6.cfg and see if that now works for you?
Barry
Barry
The option list still assumes that HDF5 has been installed manually into /opt/hdf5/1.8.7. If this option list is supposed to be easy to use, I would remove this, and let Cactus build HDF5.
-erik
On Tue, Dec 11, 2012 at 1:56 PM, Barry Wardell barry.wardell@gmail.comwrote:
On Mon, Dec 10, 2012 at 4:59 PM, Frank Loeffler knarf@cct.lsu.edu wrote:
On Mon, Dec 10, 2012 at 03:56:08PM -0600, Frank Loeffler wrote:
Try installing the gsl-devel package. This should have been caught by the GSL configuration script though.
Having said that, sl6.cfg specifies a GSL dir (and sets it to /usr). I don't think this is a good idea. If gsl is installed in /usr the configure script should be able to find it (and catch your problem) and then build the included gsl version. Should this definition be removed or is there a reason for it to be there?
You're right. It's probably better to not have it in there in case the user doesn't have the gsl-devel package installed. I have removed this (and some other similarly unnecessary options) from the optionlist. I also fixed a bug in the optionlist which prevented it from working with fortran code.
Joe, can you try the latest sl6.cfg and see if that now works for you?
Barry
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
On Tue, Dec 11, 2012 at 2:08 PM, Erik Schnetter schnetter@cct.lsu.eduwrote:
The option list still assumes that HDF5 has been installed manually into /opt/hdf5/1.8.7. If this option list is supposed to be easy to use, I would remove this, and let Cactus build HDF5.
It is intended to be useful as a general Scientific Linux 6 optionlist, but it is currently only used by one machine (tesla). We could create a separate optionlist for that machine and make sl6.cfg more generic as you suggest, but then there would be nobody using sl6.cfg and it could easily go out of date.
In the case of the Mac OS X optionlists, we decided not to do this and to just require that certain packages are installed. How does this differ from that case?
On Wed, Dec 12, 2012 at 12:56:56AM -0500, Barry Wardell wrote:
In the case of the Mac OS X optionlists, we decided not to do this and to just require that certain packages are installed. How does this differ from that case?
In theory (and we should make this work), packages from major vendors should be found automatically. Optionlists should still suggest to install them, but it should work without as well.
This currently works for, e.g., the debian option list. It doesn't mention most libraries at all, except in the comment section. If these are installed they will be found - if they are not they will be built.
Frank
The difference here is that HDF5 is expected to be installed manually, which is much more work than e.g. "sudo apt-get install gsl".
The starting point of the discussion was that there was another person using SL6, and problems ensued because GSL was expected to be installed, but this was not documented in the option list.
-erik
On Wed, Dec 12, 2012 at 12:56 AM, Barry Wardell barry.wardell@gmail.comwrote:
On Tue, Dec 11, 2012 at 2:08 PM, Erik Schnetter schnetter@cct.lsu.eduwrote:
The option list still assumes that HDF5 has been installed manually into /opt/hdf5/1.8.7. If this option list is supposed to be easy to use, I would remove this, and let Cactus build HDF5.
It is intended to be useful as a general Scientific Linux 6 optionlist, but it is currently only used by one machine (tesla). We could create a separate optionlist for that machine and make sl6.cfg more generic as you suggest, but then there would be nobody using sl6.cfg and it could easily go out of date.
In the case of the Mac OS X optionlists, we decided not to do this and to just require that certain packages are installed. How does this differ from that case?
On Wed, Dec 12, 2012 at 10:10 AM, Erik Schnetter schnetter@cct.lsu.eduwrote:
The difference here is that HDF5 is expected to be installed manually, which is much more work than e.g. "sudo apt-get install gsl".
The starting point of the discussion was that there was another person using SL6, and problems ensued because GSL was expected to be installed, but this was not documented in the option list.
OK, I have made the SL6 optionlist as generic as possible and included the relevant 'yum install gsl-devel' comments. If there are packages which I forgot to mention, please let me know.
I installed gsl using yum as indicated below. (Note: I didn't modify sl6.cfg at all)
I'm not getting the error I was, but then I got
Creating hwloc-gather-topology.1 man page... CCLD lstopo-no-graphics CCLD hwloc-assembler /usr/bin/ld:uca/bin/ld: cannot find -lzc collect2: ld returned 1 exit status make[3]: *** [lstopo-no-graphics] Error 1 make[3]: *** Waiting for unfinished jobs.... make[3]: *** [hwloc-assembler] Error 1 make[2]: *** [all-recursive] Error 1
CST error 1: -> Configuration script for thorn HWLOC returned exit code 1 (no error message)
hwloc selected, but HWLOC_DIR not set. Checking some places...
I was able to get past that by installing hwloc-devel using yum as well.
Now I'm getting some sort of complile error
COMPILING /home/hellmersjl/EinsteinToolkit/Cactus/arrangements/EinsteinInitialData/Hydro_InitExcision/src/Hydro_InitExcision.c make[3]: *** [announce.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/lib/libthorn_Formaline.a] Error 2 make[1]: *** Waiting for unfinished jobs.... COMPILING /home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/bindings/build/Hydro_InitExcision/cctk_ThornBindings.c Creating /home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/lib/libthorn_Hydro_InitExcision.a make: *** [sim] Error 2
Not sure what to try next.
Joe
On Dec 12, 2012, at 8:04 PM, Barry Wardell wrote:
On Wed, Dec 12, 2012 at 10:10 AM, Erik Schnetter schnetter@cct.lsu.edu wrote: The difference here is that HDF5 is expected to be installed manually, which is much more work than e.g. "sudo apt-get install gsl".
The starting point of the discussion was that there was another person using SL6, and problems ensued because GSL was expected to be installed, but this was not documented in the option list.
OK, I have made the SL6 optionlist as generic as possible and included the relevant 'yum install gsl-devel' comments. If there are packages which I forgot to mention, please let me know.
Joe
You are probably using a parallel make that runs several commands at once. The failing command is thus not always the last one before the error message.
Could you show us a bit more context, e.g. the twenty lines before the error message?
-erik
On Thu, Dec 13, 2012 at 8:04 PM, Joe Hellmers jhellmer@rohan.sdsu.eduwrote:
I installed gsl using yum as indicated below. (Note: I didn't modify sl6.cfg at all)
I'm not getting the error I was, but then I got
Creating hwloc-gather-topology.1 man page... CCLD lstopo-no-graphics CCLD hwloc-assembler /usr/bin/ld:uca/bin/ld: cannot find -lzc collect2: ld returned 1 exit status make[3]: *** [lstopo-no-graphics] Error 1 make[3]: *** Waiting for unfinished jobs.... make[3]: *** [hwloc-assembler] Error 1 make[2]: *** [all-recursive] Error 1
CST error 1: -> Configuration script for thorn HWLOC returned exit code 1 (no error message)
hwloc selected, but HWLOC_DIR not set. Checking some places...
I was able to get past that by installing hwloc-devel using yum as well.
Now I'm getting some sort of complile error
COMPILING /home/hellmersjl/EinsteinToolkit/Cactus/arrangements/EinsteinInitialData/Hydro_InitExcision/src/Hydro_InitExcision.c make[3]: *** [announce.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/lib/libthorn_Formaline.a] Error 2 make[1]: *** Waiting for unfinished jobs.... COMPILING /home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/bindings/build/Hydro_InitExcision/cctk_ThornBindings.c Creating /home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/lib/libthorn_Hydro_InitExcision.a make: *** [sim] Error 2
Not sure what to try next.
Joe
On Dec 12, 2012, at 8:04 PM, Barry Wardell wrote:
On Wed, Dec 12, 2012 at 10:10 AM, Erik Schnetter schnetter@cct.lsu.eduwrote:
The difference here is that HDF5 is expected to be installed manually, which is much more work than e.g. "sudo apt-get install gsl".
The starting point of the discussion was that there was another person using SL6, and problems ensued because GSL was expected to be installed, but this was not documented in the option list.
OK, I have made the SL6 optionlist as generic as possible and included the relevant 'yum install gsl-devel' comments. If there are packages which I forgot to mention, please let me know.
Attached is the complete log from the build.
On Dec 13, 2012, at 6:32 PM, Erik Schnetter wrote:
Joe
You are probably using a parallel make that runs several commands at once. The failing command is thus not always the last one before the error message.
Could you show us a bit more context, e.g. the twenty lines before the error message?
-erik
On Thu, Dec 13, 2012 at 8:04 PM, Joe Hellmers jhellmer@rohan.sdsu.edu wrote: I installed gsl using yum as indicated below. (Note: I didn't modify sl6.cfg at all)
I'm not getting the error I was, but then I got
Creating hwloc-gather-topology.1 man page... CCLD lstopo-no-graphics CCLD hwloc-assembler /usr/bin/ld:uca/bin/ld: cannot find -lzc collect2: ld returned 1 exit status make[3]: *** [lstopo-no-graphics] Error 1 make[3]: *** Waiting for unfinished jobs.... make[3]: *** [hwloc-assembler] Error 1 make[2]: *** [all-recursive] Error 1
CST error 1: -> Configuration script for thorn HWLOC returned exit code 1 (no error message)
hwloc selected, but HWLOC_DIR not set. Checking some places...
I was able to get past that by installing hwloc-devel using yum as well.
Now I'm getting some sort of complile error
COMPILING /home/hellmersjl/EinsteinToolkit/Cactus/arrangements/EinsteinInitialData/Hydro_InitExcision/src/Hydro_InitExcision.c make[3]: *** [announce.cc.o] Error 1 make[2]: *** [make.checked] Error 2 make[1]: *** [/home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/lib/libthorn_Formaline.a] Error 2 make[1]: *** Waiting for unfinished jobs.... COMPILING /home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/bindings/build/Hydro_InitExcision/cctk_ThornBindings.c Creating /home/hellmersjl/EinsteinToolkit/Cactus/configs/sim/lib/libthorn_Hydro_InitExcision.a make: *** [sim] Error 2
Not sure what to try next.
Joe
On Dec 12, 2012, at 8:04 PM, Barry Wardell wrote:
On Wed, Dec 12, 2012 at 10:10 AM, Erik Schnetter schnetter@cct.lsu.edu wrote: The difference here is that HDF5 is expected to be installed manually, which is much more work than e.g. "sudo apt-get install gsl".
The starting point of the discussion was that there was another person using SL6, and problems ensued because GSL was expected to be installed, but this was not documented in the option list.
OK, I have made the SL6 optionlist as generic as possible and included the relevant 'yum install gsl-devel' comments. If there are packages which I forgot to mention, please let me know.
-- Erik Schnetter schnetter@cct.lsu.edu http://www.perimeterinstitute.ca/personal/eschnetter/
Hi Joe,
It looks like you are missing OpenMPI.
Are you using the very latest version of sl6.cfg? That's the one here: https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/sl6.c.... At the top of that it gives commands for installing gsl-devel and openmpi-devel. The second of these should provide the necessary OpenMPI.
hwloc selected, but HWLOC_DIR not set. Checking some places...
I was able to get past that by installing hwloc-devel using yum as well.
Unfortunately, the version provided by hwloc-devel is too old. If you use the latest sl6.cfg it will build hwloc instead.
Barry
That's seems to help but now I seem to be getting an error on FFTW3
FFTW3 Implements: FFTW3 FFTW3 FFTW3 FFTW3: Preparing directory structure... FFTW3: Unpacking archive... ~/EinsteinToolkit/Cactus/configs/sim/scratch/build/FFTW3 ~/EinsteinToolkit/Cactus/configs/sim/scratch FFTW3: Configuring... FFTW3: Building... -> Configuration script for thorn FFTW3 returned exit code 1 FFTW3 selected, but FFTW3_DIR not set. Checking some places... FFTW3 not found Using bundled FFTW3... Building FFTW3 patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib_g -> Configuration script for thorn FFTW3 returned exit code 1
Is this another thing I need to install?
Full log is attached.
On Dec 13, 2012, at 10:12 PM, Barry Wardell wrote:
Hi Joe,
It looks like you are missing OpenMPI.
Are you using the very latest version of sl6.cfg? That's the one here: https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/sl6.c.... At the top of that it gives commands for installing gsl-devel and openmpi-devel. The second of these should provide the necessary OpenMPI.
hwloc selected, but HWLOC_DIR not set. Checking some places...
I was able to get past that by installing hwloc-devel using yum as well.
Unfortunately, the version provided by hwloc-devel is too old. If you use the latest sl6.cfg it will build hwloc instead.
Barry
Can you try installing the fftw-devel package? If that fixes it, I'll add that to the list in sl6.cfg.
Barry
On Fri, Dec 14, 2012 at 10:17 AM, Joe Hellmers jhellmer@rohan.sdsu.eduwrote:
That's seems to help but now I seem to be getting an error on FFTW3
FFTW3 Implements: FFTW3 FFTW3 FFTW3 FFTW3: Preparing directory structure... FFTW3: Unpacking archive... ~/EinsteinToolkit/Cactus/configs/sim/scratch/build/FFTW3 ~/EinsteinToolkit/Cactus/configs/sim/scratch FFTW3: Configuring... FFTW3: Building... -> Configuration script for thorn FFTW3 returned exit code 1 FFTW3 selected, but FFTW3_DIR not set. Checking some places... FFTW3 not found Using bundled FFTW3... Building FFTW3 patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib_g -> Configuration script for thorn FFTW3 returned exit code 1
Is this another thing I need to install?
Full log is attached.
On Dec 13, 2012, at 10:12 PM, Barry Wardell wrote:
Hi Joe,
It looks like you are missing OpenMPI.
Are you using the very latest version of sl6.cfg? That's the one here: https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/sl6.c.... At the top of that it gives commands for installing gsl-devel and openmpi-devel. The second of these should provide the necessary OpenMPI.
hwloc selected, but HWLOC_DIR not set. Checking some places...
I was able to get past that by installing hwloc-devel using yum as well.
Unfortunately, the version provided by hwloc-devel is too old. If you use the latest sl6.cfg it will build hwloc instead.
Barry
That seemed to fix that.
Now I'm getting an error that mpi.h can't be found.
Is MPI something I can install with yum too?
Joe
On Dec 14, 2012, at 9:05 AM, Barry Wardell wrote:
Can you try installing the fftw-devel package? If that fixes it, I'll add that to the list in sl6.cfg.
Barry
On Fri, Dec 14, 2012 at 10:17 AM, Joe Hellmers jhellmer@rohan.sdsu.edu wrote: That's seems to help but now I seem to be getting an error on FFTW3
FFTW3 Implements: FFTW3 FFTW3 FFTW3 FFTW3: Preparing directory structure... FFTW3: Unpacking archive... ~/EinsteinToolkit/Cactus/configs/sim/scratch/build/FFTW3 ~/EinsteinToolkit/Cactus/configs/sim/scratch FFTW3: Configuring... FFTW3: Building... -> Configuration script for thorn FFTW3 returned exit code 1 FFTW3 selected, but FFTW3_DIR not set. Checking some places... FFTW3 not found Using bundled FFTW3... Building FFTW3 patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib_g -> Configuration script for thorn FFTW3 returned exit code 1
Is this another thing I need to install?
Full log is attached.
On Dec 13, 2012, at 10:12 PM, Barry Wardell wrote:
Hi Joe,
It looks like you are missing OpenMPI.
Are you using the very latest version of sl6.cfg? That's the one here: https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/sl6.c.... At the top of that it gives commands for installing gsl-devel and openmpi-devel. The second of these should provide the necessary OpenMPI.
hwloc selected, but HWLOC_DIR not set. Checking some places...
I was able to get past that by installing hwloc-devel using yum as well.
Unfortunately, the version provided by hwloc-devel is too old. If you use the latest sl6.cfg it will build hwloc instead.
Barry
This should have been provided by the openmpi-devel package. Have you installed that? On 14 Dec 2012 13:04, "Joe Hellmers" jhellmer@rohan.sdsu.edu wrote:
That seemed to fix that.
Now I'm getting an error that mpi.h can't be found.
Is MPI something I can install with yum too?
Joe
On Dec 14, 2012, at 9:05 AM, Barry Wardell wrote:
Can you try installing the fftw-devel package? If that fixes it, I'll add that to the list in sl6.cfg.
Barry
On Fri, Dec 14, 2012 at 10:17 AM, Joe Hellmers jhellmer@rohan.sdsu.eduwrote:
That's seems to help but now I seem to be getting an error on FFTW3
FFTW3 Implements: FFTW3 FFTW3 FFTW3 FFTW3: Preparing directory structure... FFTW3: Unpacking archive... ~/EinsteinToolkit/Cactus/configs/sim/scratch/build/FFTW3 ~/EinsteinToolkit/Cactus/configs/sim/scratch FFTW3: Configuring... FFTW3: Building... -> Configuration script for thorn FFTW3 returned exit code 1 FFTW3 selected, but FFTW3_DIR not set. Checking some places... FFTW3 not found Using bundled FFTW3... Building FFTW3 patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib_g -> Configuration script for thorn FFTW3 returned exit code 1
Is this another thing I need to install?
Full log is attached.
On Dec 13, 2012, at 10:12 PM, Barry Wardell wrote:
Hi Joe,
It looks like you are missing OpenMPI.
Are you using the very latest version of sl6.cfg? That's the one here: https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/sl6.c.... At the top of that it gives commands for installing gsl-devel and openmpi-devel. The second of these should provide the necessary OpenMPI.
hwloc selected, but HWLOC_DIR not set. Checking some places...
I was able to get past that by installing hwloc-devel using yum as well.
Unfortunately, the version provided by hwloc-devel is too old. If you use the latest sl6.cfg it will build hwloc instead.
Barry
I installed openmpi and now it looks like everything worked OK.
Tons of warning messages though
Joe
On Dec 14, 2012, at 9:05 AM, Barry Wardell wrote:
Can you try installing the fftw-devel package? If that fixes it, I'll add that to the list in sl6.cfg.
Barry
On Fri, Dec 14, 2012 at 10:17 AM, Joe Hellmers jhellmer@rohan.sdsu.edu wrote: That's seems to help but now I seem to be getting an error on FFTW3
FFTW3 Implements: FFTW3 FFTW3 FFTW3 FFTW3: Preparing directory structure... FFTW3: Unpacking archive... ~/EinsteinToolkit/Cactus/configs/sim/scratch/build/FFTW3 ~/EinsteinToolkit/Cactus/configs/sim/scratch FFTW3: Configuring... FFTW3: Building... -> Configuration script for thorn FFTW3 returned exit code 1 FFTW3 selected, but FFTW3_DIR not set. Checking some places... FFTW3 not found Using bundled FFTW3... Building FFTW3 patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib patching file Lorene/C++/Source/Non_class_members/Coef/FFTW3/Makefile_lib_g -> Configuration script for thorn FFTW3 returned exit code 1
Is this another thing I need to install?
Full log is attached.
On Dec 13, 2012, at 10:12 PM, Barry Wardell wrote:
Hi Joe,
It looks like you are missing OpenMPI.
Are you using the very latest version of sl6.cfg? That's the one here: https://svn.cct.lsu.edu/repos/numrel/simfactory2/trunk/mdb/optionlists/sl6.c.... At the top of that it gives commands for installing gsl-devel and openmpi-devel. The second of these should provide the necessary OpenMPI.
hwloc selected, but HWLOC_DIR not set. Checking some places...
I was able to get past that by installing hwloc-devel using yum as well.
Unfortunately, the version provided by hwloc-devel is too old. If you use the latest sl6.cfg it will build hwloc instead.
Barry
users@lists.einsteintoolkit.org
-
Barry Wardell -
Eloisa Bentivegna -
Erik Schnetter -
Frank Loeffler -
Joe Hellmers