July 2014 - discuss - mailman3.cels.anl.gov

Problem with "sock"-channel
by Klaus Philip 08 Jul '14

08 Jul '14

After reconfiguring, recompiling and reinstalling mpich2 with the sock-channel flag (--with-device=ch3:sock) to avoid active polling, every MPI program (Self written programs, and also the Intel MPI Benchmark too) breaks when it calls a MPI function: 1) Error at my self written MPI program: [X] 2) Error at the Intel MPI Benchmark: [X] So I hope someone of you knows what issue is causing these errors. Greetings from Austria, Philip

1 0

Re: [mpich-discuss] couldn't build mpich-master-v3.1.1-56-g9d5ce23c on Solaris 10 x86_64 with gcc-4.9.0
by Rob Latham 07 Jul '14

07 Jul '14

On 07/06/2014 04:52 AM, Siegmar Gross wrote: > > > Next I checked my file "log.make.SunOS.x86_64.64_gcc" and got the > following error. > > ... > CC adio/common/ad_prealloc.lo > CC adio/common/ad_read.lo > In file included from /usr/include/unistd.h:18:0, > from > ../../../../mpich-master-v3.1.1-56-g9d5ce23c/src/mpi/romio/adio/common/ad_read.c > :10: > /export2/prog/SunOS_x86_64/gcc-4.9.0/lib/gcc/i386-pc-solaris2.10/4.9.0/include-f > ixed/sys/feature_tests.h:350:2: > error: #error "Compiler or options invalid; UNIX 03 and POSIX.1-2001 > applications require the use of c99" > #error "Compiler or options invalid; UNIX 03 and POSIX.1-2001 applications \ > ^ > make[3]: *** [adio/common/ad_read.lo] Error 1 As for this one problem, Ken and I have been working on a patch set to fix this issue. > I solved this problem setting CFLAGS to "-m64 -std=c11". Starting > once more with an empty directory, I got the same error as with > mpich-3.1.1. > > ... > CXX src/binding/cxx/initcxx.lo > CXXLD lib/libmpicxx.la > ld: fatal: file libmpicxx.so.12: open failed: No such file or directory > ld: fatal: file processing errors. No output written to > lib/.libs/libmpicxx.so.12.0.1 > collect2: error: ld returned 1 exit status > make[2]: *** [lib/libmpicxx.la] Error 1 > ... > Sounds like we'll run into this one next. Thanks for the heads-up. Not sure (yet!) how to fix it. We relatively recently added solaris to our buildbot, so after these rounds of porting issues we should stop regressing. ==rob -- Rob Latham Mathematics and Computer Science Division Argonne National Lab, IL USA

1 0

couldn't build mpich-master-v3.1.1-56-g9d5ce23c on Solaris 10 x86_64 with gcc-4.9.0
by Siegmar Gross 06 Jul '14

06 Jul '14

Hi, today I tried to build "mpich-master-v3.1.1-56-g9d5ce23c" on "Solaris 10 x86_64" with gcc-4.9.0 in a 64-bit version. I was able to configure the package with some warnings, but unable to build it. First I checked my file "log.configure.SunOS.x86_64.64_gcc" for warnings. Some warnings can be avoided, if somebody has time to change a few things in the configure script. configure: WARNING: Unknown architecture i86pc... proceeding anyway You can use "uname -p" or "uname -m" and then "isainfo -k" to check if you build the package on a 32-bit or 64-bit processor for "Solaris x86_64" . sunpc1 fd1026 108 uname -s SunOS sunpc1 fd1026 109 uname -p i386 sunpc1 fd1026 110 uname -m i86pc sunpc1 fd1026 103 isainfo -k amd64 I use the following code in the initialization script for my heterogenous environment to determine the operating system and processor architecture. setenv SYSTEM_ENV `uname -s` if (${SYSTEM_ENV} == Linux) then setenv MACHINE_ENV `uname -m` else setenv MACHINE_ENV `uname -p` if ( (${SYSTEM_ENV} == SunOS) && (${MACHINE_ENV} != sparc) ) then if (`isainfo -k` == amd64) then setenv MACHINE_ENV "x86_64" endif endif endif configure: WARNING: X11/Xutil.h: present but cannot be compiled configure: WARNING: X11/Xutil.h: check for missing prerequisite headers? configure: WARNING: X11/Xutil.h: see the Autoconf documentation configure: WARNING: X11/Xutil.h: section "Present But Cannot Be Compiled" configure: WARNING: X11/Xutil.h: proceeding with the compiler's result You can compile "X11/Xutil.h" on Solaris if you include <X11/Xlib.h> before including "X11/Xutil.h". configure: WARNING: The ch3 device did not set the maximum size of an error string, 512 being used. config.status: WARNING: '../mpich-master-v3.1.1-56-g9d5ce23c/src/packaging/envmods/mpich.module.in' seems to ignore the --datarootdir setting I don't know if the above warnings are important or not. Next I checked my file "log.make.SunOS.x86_64.64_gcc" and got the following error. ... CC adio/common/ad_prealloc.lo CC adio/common/ad_read.lo In file included from /usr/include/unistd.h:18:0, from ../../../../mpich-master-v3.1.1-56-g9d5ce23c/src/mpi/romio/adio/common/ad_read.c :10: /export2/prog/SunOS_x86_64/gcc-4.9.0/lib/gcc/i386-pc-solaris2.10/4.9.0/include-f ixed/sys/feature_tests.h:350:2: error: #error "Compiler or options invalid; UNIX 03 and POSIX.1-2001 applications require the use of c99" #error "Compiler or options invalid; UNIX 03 and POSIX.1-2001 applications \ ^ make[3]: *** [adio/common/ad_read.lo] Error 1 ... I solved this problem setting CFLAGS to "-m64 -std=c11". Starting once more with an empty directory, I got the same error as with mpich-3.1.1. ... CXX src/binding/cxx/initcxx.lo CXXLD lib/libmpicxx.la ld: fatal: file libmpicxx.so.12: open failed: No such file or directory ld: fatal: file processing errors. No output written to lib/.libs/libmpicxx.so.12.0.1 collect2: error: ld returned 1 exit status make[2]: *** [lib/libmpicxx.la] Error 1 ... I would be grateful if somebody could provide a solution for my problem. Do you need anything else? Thank you very much for any help in advance. Kind regards Siegmar

1 0

Re: [mpich-discuss] Random Aborts in An MPI Program
by Balaji, Pavan 05 Jul '14

05 Jul '14

Mark, Sorry, 1.0.6 is too old for us to help you here. It might be a bug that was resolved a long time ago. You can install mpich in your home directory. You don’t need root permissions to install it. Please see the README for such instructions. Regards, — Pavan On Jul 5, 2014, at 3:34 PM, mark <dimitsas.markos(a)gmail.com> wrote: > Hello to all and to whoever is from the United States, Happy 4th of Jully ! > > > I am having these collective aborts while executing a program. More specifically the error is : > rank 7 in job 38 Calliope_50667 caused collective abort of all ranks exit status of rank 7: killed by signal 11 [cli_2]: aborting job: Fatal error in MPI_Allgather: Error message texts are not available [cli_4]: aborting job: Fatal error in MPI_Allgather: Error message texts are not available > > But the MPI_Allgather command it's written correctly, since i checked it multiple times and what about the "Error message texts are not available"? > > MPI_Allgather(&docs, 1, MPI_INT, texts_vectors, 1, MPI_INT, MPI_COMM_WORLD); > > Where docs is a int variable holding a single value for each node, and texts_vectors is an int array with the size of the population of the nodes. > > I compile the programm using mpicc -g -o prog prog.c -lm and execute using mpiexec -n number_of_nodes `pwd`/prog > I am using MPICH2 1.0.6 in a linux cluster machine that i use for the purposes of my bachelor thesis. I know it's an older version, but the machine belongs to an institute, so i don't have the permission to upgrade it. > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss

1 0

Random Aborts in An MPI Program
by mark 05 Jul '14

05 Jul '14

Hello to all and to whoever is from the United States, /Happy 4th of Jully/ ! I am having these collective aborts while executing a program. More specifically the error is : /rank 7 in job 38 Calliope_50667 caused collective abort of all ranks exit status of rank 7: killed by signal 11 [cli_2]: aborting job: Fatal error in MPI_Allgather: Error message texts are not available [cli_4]: aborting job: Fatal error in MPI_Allgather: Error message texts are not available// / But the /MPI_Allgather/ command it's written correctly, since i checked it multiple times and what about the "/Error message texts are not available/"? /MPI_Allgather(&docs, 1, MPI_INT, texts_vectors, 1, MPI_INT, MPI_COMM_WORLD);/// Where /docs/ is a int variable holding a single value for each node, and /texts_vectors/ is an int array with the size of the population of the nodes. I compile the programm using /mpicc -g -o prog prog.c -lm/ and execute using /mpiexec -n number_of_nodes `pwd`/prog/ I am using MPICH2 1.0.6 in a linux cluster machine that i use for the purposes of my bachelor thesis. I know it's an older version, but the machine belongs to an institute, so i don't have the permission to upgrade it.

1 0

Re: [mpich-discuss] FW: Problem validating MPICH installation for Windows 7
by Bland, Wesley B. 03 Jul '14

03 Jul '14

Joaquin, Sorry but we haven’t provided support for Windows installations for quite some time down. We don’t have the resources or personnel to do so. However, if you would like to use MPI on Windows, we do suggest using either the Microsoft version which is available for free or the Intel version which is paid. Both are derivatives of MPICH which have their own additions to work on specific platforms or hardware. You can find more information at the downloads page of MPICH (www.mpich.org/downloads<http://www.mpich.org/downloads>). Thanks, Wesley On Jul 2, 2014, at 11:19 PM, Noyola, Joaquin P <joaquin.noyola(a)mavs.uta.edu<mailto:[email protected]>> wrote: Dear MPICH.org<http://mpich.org/>, I recently tried to install MPICH2 on my Windows 7 64-bit machine using your binary version 1.4.1p1, and I had a small problem on the very last step (I followed the instructions given in http://swash.sourceforge.net/online_doc/swashimp/node9.html) After installing the binary, I went to the command prompt and entered the following: smpd -install mpiexec -remove mpiexec -register mpiexec -validate smpd -status Every step worked fine, even changing the pass phrase, but the mpiexec -validate command returns FAIL. I tried reinstalling the package again, but I got the same result. My question is: is there a way to get more information about why the validation failed? Thank you for your time, I look forward to your response. Sincerely, Joaquin P Noyola Graduate Teaching Assistant Department of Physics, UT Arlington _______________________________________________ discuss mailing list discuss(a)mpich.org<mailto:[email protected]> To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

FW: Problem validating MPICH installation for Windows 7
by Noyola, Joaquin P 02 Jul '14

02 Jul '14

Dear MPICH.org, I recently tried to install MPICH2 on my Windows 7 64-bit machine using your binary version 1.4.1p1, and I had a small problem on the very last step (I followed the instructions given in http://swash.sourceforge.net/online_doc/swashimp/node9.html?) After installing the binary, I went to the command prompt and entered the following: smpd -install mpiexec -remove mpiexec -register mpiexec -validate smpd -status Every step worked fine, even changing the pass phrase, but the mpiexec -validate command returns FAIL. I tried reinstalling the package again, but I got the same result. My question is: is there a way to get more information about why the validation failed?? Thank you for your time, I look forward to your response. Sincerely, Joaquin P Noyola Graduate Teaching Assistant Department of Physics, UT Arlington

1 0

Re: [mpich-discuss] mpicc shell error of some kind
by Brian Van Straalen 02 Jul '14

02 Jul '14

I honestly don’t know how GCC is finding openmpi. Yes, my Ubuntu ships with OpenMPI on it, but how is my mpich mpicc shell script removing mpich from the compile command? hmmm, unless make is not taking my environment with it when sub make commands for some of my packages….. I get mpicc from mpich-3.1.1 through my modules, but openmpi is installed in the default /usr/local location…. I think I see the drift of the problem. On Jul 2, 2014, at 5:33 PM, Dave Goodell (dgoodell) <dgoodell(a)cisco.com> wrote: > On Jul 2, 2014, at 3:06 PM, Brian Van Straalen <bvstraalen(a)lbl.gov> wrote: > >> >> an interesting error while using mpicc with mpich-3.1.1 >> >>> which mpicc >> /opt/pkg/mpich-3.1.1/bin/mpicc >>> mpicc -compile-info >> gcc -I/opt/pkg/mpich-3.1.1/include -L/opt/pkg/mpich-3.1.1/lib -Wl,-rpath > [...] >> '/usr/lib/openmpi/include' '-I' '/usr/lib/openmpi/include/openmpi' '-pthread' > > That line looks awfully suspicious to me. You probably have a conflicting installation of Open MPI. Maybe this is an older version of OS X that came with Open MPI installed in /usr? > > -Dave > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss Brian Van Straalen Lawrence Berkeley Lab BVStraalen(a)lbl.gov Computational Research (510) 486-4976 Division (crd.lbl.gov)

1 0

Re: [mpich-discuss] mpicc shell error of some kind
by Dave Goodell (dgoodell) 02 Jul '14

02 Jul '14

On Jul 2, 2014, at 3:06 PM, Brian Van Straalen <bvstraalen(a)lbl.gov> wrote: > > an interesting error while using mpicc with mpich-3.1.1 > > >which mpicc > /opt/pkg/mpich-3.1.1/bin/mpicc > > mpicc -compile-info > gcc -I/opt/pkg/mpich-3.1.1/include -L/opt/pkg/mpich-3.1.1/lib -Wl,-rpath [...] > '/usr/lib/openmpi/include' '-I' '/usr/lib/openmpi/include/openmpi' '-pthread' That line looks awfully suspicious to me. You probably have a conflicting installation of Open MPI. Maybe this is an older version of OS X that came with Open MPI installed in /usr? -Dave

1 0

Re: [mpich-discuss] Possible bug MPICH2 1.0.6
by Junchao Zhang 02 Jul '14

02 Jul '14

It is very likely a bug in your code. Without a toy program to reproduce your error, it is hard for us to help. Also, could you use newer MPICH releases like MPICH 3.1.1? --Junchao Zhang On Wed, Jul 2, 2014 at 3:02 PM, mark <dimitsas.markos(a)gmail.com> wrote: > Hello. > In a program i am writing, i have an array which size is equal to the > number of nodes in my cluster. > I made this array to keep an eye of the objects that each node has in his > control and when i use it with the actual ranks of the nodes as positions ( > *array[node_rank]=objects;*) it returns me an error and stops the > execution: > > *rank 0 in job 10 Calliope_49755 caused collective abort of all ranks* > * exit status of rank 0: killed by signal 11 * > > However, i had freely used the variable *id* or *node_rank* to point out > a specific action for a specific node since now and never had problems. > > If instead of using the ranks node as a pointer for the array, i use an > integer i.e. 5, the program runs fine, but even now, 1 out of 7 executions, > it returns me an error and stops. The action i am calling that keeps > returning me the error is a simple abstraction (*array[id]--;*). > > Any ideas? > > > PS. The array is 1d and is created by allocating memory, like this : *int > *array = malloc(processes * sizeof(int));* > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss >

1 0