You are welcome. Have no idea how that came about there... Glad it was fixed! :)
Kind regards Nick 2011/10/5 David Noriega <[email protected]> > Ah HA! that was it. I wonder how I did that, but that fixed it. Thanks > > David > > On Wed, Oct 5, 2011 at 1:49 PM, Nick Papior Andersen > <[email protected]> wrote: > > I can see that your Interfaces.f90 is compiled before the kinds are > created. > > That should not occur. After all it is the compilation of that file that > > generates the module files that is requested. It seems as if your > computer > > setup is wrong on some parts, or at least the Make script is handled > > incorrectly? > > It compiles an Interfaces.f90 in your source folder. It should generate > one > > locally. Try and move > > /root/workspace/siesta-3.1/Src/MPI/Interfaces.f90 > > to > > /root/workspace/siesta-3.1/Src/MPI/Interfaces_old.f90 > > And try compilation... > > > > 2011/10/5 David Noriega <[email protected]> > >> > >> Yea I figured changing the code wouldnt be a good idea. > >> > >> So I started fresh, even stuck with using Obj instead of some other > >> folder name, ran obj_setup.sh, rand configure --enable-mpi(didn't > >> bother with math libraries since it will crap out before it even > >> bothers with linking those) and get the same error. > >> > >> make[1]: Entering directory `/root/workspace/siesta-3.1/Obj/MPI' > >> mpif90 -c -g -O2 -I. > /root/workspace/siesta-3.1/Src/MPI/mpi__include.f90 > >> mpif90 -c -g -O2 -I. /root/workspace/siesta-3.1/Src/MPI/Interfaces.f90 > >> if [ -z "4 8" ] ; then make kind_explorer ; fi > >> The kind numbers for single and double precision reals follow > >> sh /root/workspace/siesta-3.1/Src/MPI/generate.sh "4 8" > >> ===> Generating module files from templates... > >> 4 8 > >> mpif90 -c -g -O2 -I. -DMPI -DFC_HAVE_FLUSH -DFC_HAVE_ABORT > >> /root/workspace/siesta-3.1/Src/MPI/mpi.F > >> V_S.uses:1.29: > >> Included at /root/workspace/siesta-3.1/Src/MPI/mpi.F:54: > >> > >> USE MPI__r4_V ; USE MPI__r4_S > >> 1 > >> Fatal Error: Can't open module file 'mpi__r4_v.mod' for reading at > >> (1): No such file or directory > >> make[1]: *** [mpi.o] Error 1 > >> make[1]: Leaving directory `/root/workspace/siesta-3.1/Obj/MPI' > >> make: *** [libmpi_f90.a] Error 2 > >> > >> I understand others have built it with no problem, but I don't get why > >> I'm getting this problem. My guess is they files are generated at some > >> point, but why are they not being generated for me? My arch.make is > >> posted in the first message and from what I can tell, its all good. So > >> whats up? > >> > >> Thanks > >> David > >> ---------- Forwarded message ---------- > >> From: Alberto Garcia <[email protected]> > >> Date: Wed, Oct 5, 2011 at 12:24 PM > >> Subject: Re: [SIESTA-L] MPI Compile Error > >> To: [email protected] > >> > >> > >> Dear all, > >> > >> There should be no need to change anything in the code. Make sure you > >> understand the issues involved in compilation (see for example > >> the example, documented .make file in Src/Sys, and the sample files > >> there), create the appropriate arch.make for you, and maybe start over > >> from a clean working directory. Many users compile Siesta in parallel > >> routinely. It is not completely trivial, but not rocket science (maybe > >> materials science... :-) ). > >> > >> Alberto > >> > >> > >> > >> On Wed, Oct 5, 2011 at 6:47 PM, Ray Sheppard <[email protected]> > wrote: > >> > Hello David and everyone, > >> > There are many possible issues (like consistancy with the libraries > and > >> > the > >> > build), but > >> > if the trouble is just the MPI mod files, you can search for "USE MPI" > >> > lines > >> > in the > >> > source code and replace them with "include mpif.h" directives. > >> > Ray > >> > > >> > On 10/5/2011 11:03 AM, David Noriega wrote: > >> >> > >> >> yes openmpi works, I've built other programs against it like abinit > >> >> and many others. And yes thats what I've read that mpif90 just takes > >> >> care of that, but adding the include path was just something to try > to > >> >> figure out whats going wrong. What is this mod file and why is it > >> >> missing? > >> >> > >> >> I've tried to build it against mpich2 and again the error with the > >> >> missing mod file. > >> >> > >> >> The "parallel" directory was just to keep the builds separate, and > yes > >> >> I know the parallel build can run in serial mode, but with this > >> >> problem, cant build it, but yes the serial version builds just fine. > >> >> > >> >> On Wed, Oct 5, 2011 at 6:41 AM, RCP<[email protected]> wrote: > >> >>> > >> >>> Does openmpi work correctly for you ?, have you tried it before ?. > >> >>> There > >> >>> should > >> >>> be no need of that " /opt/openmpi/include " , because mpif90 already > >> >>> takes > >> >>> care. > >> >>> > >> >>> Also, I see that "parallel" in your path; you changed it from the > >> >>> standard > >> >>> "Obj", > >> >>> right ?. Besides, you know that a parallel binary can be run in > serial > >> >>> mode > >> >>> without > >> >>> performance loss, don't you ? (I recall you said to have gotten the > >> >>> serial > >> >>> stuff ok). > >> >>> > >> >>> R. > >> >>> > >> >>> > >> >>> David Noriega wrote: > >> >>>> > >> >>>> I did figure this much out. If I leave MPI_INCLUDE=. I get the > error > >> >>>> as before, but if I change MPI_INCLUDE=/opt/openmpi/include > >> >>>> I get this error: > >> >>>> make[1]: Entering directory > `/root/workspace/siesta-3.1/parallel/MPI' > >> >>>> /opt/openmpi/bin/mpif90 -c -g -O3 -I/opt/openmpi/include > >> >>>> -DFC_HAVE_FLUSH -DFC_HAVE_ABORT -DGFORTRAN -DGRID_DP -DPHI_GRID_SP > >> >>>> -DMPI /root/workspace/siesta-3.1/Src/MPI/mpi.F > >> >>>> Error: Can't open included file 'V_S.uses' > >> >>>> Error: Can't open included file 'VS.uses' > >> >>>> make[1]: *** [mpi.o] Error 1 > >> >>>> make[1]: Leaving directory > `/root/workspace/siesta-3.1/parallel/MPI' > >> >>>> make: *** [libmpi_f90.a] Error 2 > >> >>>> > >> >>>> If I get MPI_INCLUDE=/opt/openmpi/include -I. > >> >>>> to use both directories, I get the error as before where it cant > find > >> >>>> this mpi__r4_v.mod file > >> >>>> > >> >>>> > >> >> > >> >> > >> > > >> > -- > >> > Respectfully, > >> > Ray Sheppard > >> > [email protected] > >> > http://rc.uits.iu.edu/hpa/ > >> > 317-274-0016 > >> > > >> > Principal Analyst > >> > High Performance Applications > >> > Research Technologies > >> > University Information Technological Services > >> > IUPUI campus > >> > Indiana University > >> > > >> > My "pithy" saying: Science is the art of translating the world > >> > into language. Unfortunately, that language is mathematics. > >> > Bumper sticker wisdom: Make it idiot-proof and they will make a > >> > better idiot. > >> > > >> > > >> > >> > >> > >> -- > >> Personally, I liked the university. They gave us money and facilities, > >> we didn't have to produce anything! You've never been out of college! > >> You don't know what it's like out there! I've worked in the private > >> sector. They expect results. -Ray Ghostbusters > > > > > > > > -- > Personally, I liked the university. They gave us money and facilities, > we didn't have to produce anything! You've never been out of college! > You don't know what it's like out there! I've worked in the private > sector. They expect results. -Ray Ghostbusters >
