March 2015 - discuss - mailman3.cels.anl.gov

Re: [mpich-discuss] MPICH 3.2b1 packages for Ubuntu, Debian and Slackware (x86 and amd64)
by Huiwei Lu 27 Mar '15

27 Mar '15

Hi Marcelo, Thanks for your contribution to MPICH! It makes MPICH better and easier to use. You rock! -- Huiwei Lu On Fri, Mar 27, 2015 at 7:00 AM, Marcelo Soares Souza <marcelo(a)cebacad.net> wrote: > mpich-3.2b1 packages > > * Ubuntu 14.10 > > - amd64 > http://www.cebacad.net/files/mpich/ubuntu/mpich-3.2b1/ > mpich_3.2b1-1ubuntu_amd64.deb > > - x86 > http://www.cebacad.net/files/mpich/ubuntu/mpich-3.2b1/ > mpich_3.2b1-1ubuntu_i386.deb > > * Debian 8.0 (Jessie) > > - amd64 > http://www.cebacad.net/files/mpich/debian/mpich-3.2b1/ > mpich_3.2b1-1debian_amd64.deb > > - x86 > http://www.cebacad.net/files/mpich/debian/mpich-3.2b1/ > mpich_3.2b1-1debian_i386.deb > > * Slackware 14.1 > > - amd64 > http://www.cebacad.net/files/mpich/slackware/mpich-3.2b1/ > mpich-3.2b1-x86_64-goa-1.txz > > - x86 > http://www.cebacad.net/files/mpich/slackware/mpich-3.2b1/ > mpich-3.2b1-i486-goa-1.txz > > Enjoy! > > -- > Marcelo Soares Souza > https://marcelo.juntadados.org > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] Slave hosts panic after mpirun
by Huiwei Lu 27 Mar '15

27 Mar '15

Hi Vaibhav, Can you give us some details of how the slaves panic? -- Huiwei Lu On Fri, Mar 27, 2015 at 2:19 AM, Vaibhav Rekhate < rekhatevm11.comp(a)coep.ac.in> wrote: > Hello there! > I have setup a small cluster (only 3 nodes - 1 master, 2 slaves). > I was trying to run a sample program by using this command: > > $MPI_INSTALL_DIR/bin/mpirun -n 9 --machinefile machinefile ./a.out > > > Contents of machine file: > > host1:3 > host2:3 > host3:3 > > > Please help. > > Regards, > Vaibhav Rekhate > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

MPICH 3.2b1 packages for Ubuntu, Debian and Slackware (x86 and amd64)
by Marcelo Soares Souza 27 Mar '15

27 Mar '15

mpich-3.2b1 packages * Ubuntu 14.10 - amd64 http://www.cebacad.net/files/mpich/ubuntu/mpich-3.2b1/mpich_3.2b1-1ubuntu_a… - x86 http://www.cebacad.net/files/mpich/ubuntu/mpich-3.2b1/mpich_3.2b1-1ubuntu_i… * Debian 8.0 (Jessie) - amd64 http://www.cebacad.net/files/mpich/debian/mpich-3.2b1/mpich_3.2b1-1debian_a… - x86 http://www.cebacad.net/files/mpich/debian/mpich-3.2b1/mpich_3.2b1-1debian_i… * Slackware 14.1 - amd64 http://www.cebacad.net/files/mpich/slackware/mpich-3.2b1/mpich-3.2b1-x86_64… - x86 http://www.cebacad.net/files/mpich/slackware/mpich-3.2b1/mpich-3.2b1-i486-g… Enjoy! -- Marcelo Soares Souza https://marcelo.juntadados.org _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Slave hosts panic after mpirun
by Vaibhav Rekhate 27 Mar '15

27 Mar '15

Hello there! I have setup a small cluster (only 3 nodes - 1 master, 2 slaves). I was trying to run a sample program by using this command: $MPI_INSTALL_DIR/bin/mpirun -n 9 --machinefile machinefile ./a.out Contents of machine file: host1:3 host2:3 host3:3 Please help. Regards, Vaibhav Rekhate _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] [mvapich-discuss] mvapich-2.2rc2 build failure with psm/per-object thread-cs.
by Huiwei Lu 26 Mar '15

26 Mar '15

Hi Jian and Abhishek, Thanks for reporting this. When configured with --enable-thread-cs=per-object, MPID_cc_t will be defined as OPA_int_t, which can not be directly compared with 0 using "==". I will create a patch for this. Best regards, -- Huiwei Lu Postdoc Appointee Mathematics and Computer Science Division Argonne National Laboratory On Thu, Mar 26, 2015 at 4:32 PM, Jian Lin <lin.2180(a)osu.edu> wrote: > Hi, Abhishek, > > Thanks for your note. By using the full configure options you provided, > we are able to reproduce this behavior. However, MVAPICH2 2.1rc2 is > based on MPICH 3.1.4. We found that the error also exists with the > MPICH 3.1.4/3.2b1 even if using ch3:nemesis device. It will be good if > the MPICH team resolves this. I am cc'ing this note to the MPICH team. > > Hi, MPICH community, > > The forwarded compilation error reported by the user is related to > unmatched MPID_cc_t typedefs if some thread-related configure options > are used. The default build without these thread-related options does > not have any issue. Will you please take a look. Thanks! > > > On Mon, 23 Mar 2015 20:29:12 -0400 > Abhishek Kulkarni <adkulkar(a)indiana.edu> wrote: > > > Jonathan, > > > > I am using the following configure flags: > > > > ./configure --enable-threads=multiple --enable-async-progress > > --with-device=ch3:psm --enable-thread-cs=per-object CFLAGS=-O3 > > --enable-refcount=lock-free --enable-handle-allocation=tls > > --with-atomic-primitives > > > > The distribution is Ubuntu 12.04.5 LTS and the compiler version is gcc > > (Ubuntu/Linaro 4.6.4-1ubuntu1~12.04) 4.6.4. > > > > Abhishek > > > > > > On Tue, Mar 17, 2015 at 5:06 PM, Jonathan Perkins < > > perkinjo(a)cse.ohio-state.edu> wrote: > > > > > Thank you for your report. When trying to reproduce this problem > > > locally I am running into a similar but different build error. Can > > > you provide us more information on how you're building? Please let > > > us know the Linux distribution and version and vendor of the > > > compiler that you're using. > > > > > > On Tue, Mar 17, 2015 at 2:02 PM Abhishek Kulkarni > > > <adkulkar(a)indiana.edu> wrote: > > > > > >> Hello, > > >> > > >> While trying to build mvapich 2.2rc2, I run into a build failure > > >> when I use the following two flags together: --with-device=ch3:psm > > >> --enable-thread-cs=per-object > > >> > > >> Is this a known issue? Since I expect to mostly run in a hybrid > > >> (OpenMP+MPI) configuration, I am assuming that the granularity of > > >> critical sections might be critical for good performance. > > >> > > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_acc_ops.lo > > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_ops.lo > > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_reqops.lo > > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_sync.lo > > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rndv.lo > > >> src/mpid/ch3/src/ch3u_handle_recv_req.c: In function > > >> ‘MPIDI_CH3_ReqHandler_FOPComplete’: > > >> src/mpid/ch3/src/ch3u_handle_recv_req.c:963:9: warning: passing > > >> argument 4 of ‘MPIDI_CH3_iStartMsgv’ from incompatible pointer > > >> type [enabled by default] > > >> ./src/mpid/ch3/include/mpidimpl.h:1580:5: note: expected ‘struct > > >> MPID_Request **’ but argument is of type ‘struct MPID_Request *’ > > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_eager.lo > > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_eagersync.lo > > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_win_fns.lo > > >> CC src/mpid/ch3/src/lib_libmpi_la-mpid_abort.lo > > >> CC src/mpid/ch3/src/lib_libmpi_la-mpid_cancel_recv.lo > > >> CC src/mpid/ch3/src/lib_libmpi_la-mpid_cancel_send.lo > > >> src/mpid/ch3/src/ch3u_rndv.c: In function > > >> ‘MPIDI_CH3_PktHandler_RndvReqToSend’: > > >> src/mpid/ch3/src/ch3u_rndv.c:210:28: error: invalid operands to > > >> binary == (have ‘MPID_cc_t’ and ‘int’) > > >> make[2]: *** [src/mpid/ch3/src/lib_libmpi_la-ch3u_rndv.lo] Error 1 > > >> make[2]: *** Waiting for unfinished jobs.... > > >> src/mpid/ch3/src/ch3u_eager.c: In function > > >> ‘MPIDI_CH3_PktHandler_EagerSend’: > > >> src/mpid/ch3/src/ch3u_eager.c:880:28: error: invalid operands to > > >> binary == (have ‘MPID_cc_t’ and ‘int’) > > >> src/mpid/ch3/src/ch3u_eager.c: In function > > >> ‘MPIDI_CH3_PktHandler_ReadySend’: > > >> src/mpid/ch3/src/ch3u_eager.c:1022:28: error: invalid operands to > > >> binary == (have ‘MPID_cc_t’ and ‘int’) > > >> make[2]: *** [src/mpid/ch3/src/lib_libmpi_la-ch3u_eager.lo] Error 1 > > >> src/mpid/ch3/src/ch3u_eagersync.c: In function > > >> ‘MPIDI_CH3_PktHandler_EagerSyncSend’: > > >> src/mpid/ch3/src/ch3u_eagersync.c:259:28: error: invalid operands > > >> to binary == (have ‘MPID_cc_t’ and ‘int’) > > >> make[2]: *** [src/mpid/ch3/src/lib_libmpi_la-ch3u_eagersync.lo] > > >> Error 1 > > >> > > >> Thanks, > > >> Abhishek > > >> _______________________________________________ > > >> mvapich-discuss mailing list > > >> mvapich-discuss(a)cse.ohio-state.edu > > >> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss > > >> > > > > > > > -- > Jian Lin > http://linjian.org > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] [mvapich-discuss] mvapich-2.2rc2 build failure with psm/per-object thread-cs.
by Jian Lin 26 Mar '15

26 Mar '15

Hi, Abhishek, Thanks for your note. By using the full configure options you provided, we are able to reproduce this behavior. However, MVAPICH2 2.1rc2 is based on MPICH 3.1.4. We found that the error also exists with the MPICH 3.1.4/3.2b1 even if using ch3:nemesis device. It will be good if the MPICH team resolves this. I am cc'ing this note to the MPICH team. Hi, MPICH community, The forwarded compilation error reported by the user is related to unmatched MPID_cc_t typedefs if some thread-related configure options are used. The default build without these thread-related options does not have any issue. Will you please take a look. Thanks! On Mon, 23 Mar 2015 20:29:12 -0400 Abhishek Kulkarni <adkulkar(a)indiana.edu> wrote: > Jonathan, > > I am using the following configure flags: > > ./configure --enable-threads=multiple --enable-async-progress > --with-device=ch3:psm --enable-thread-cs=per-object CFLAGS=-O3 > --enable-refcount=lock-free --enable-handle-allocation=tls > --with-atomic-primitives > > The distribution is Ubuntu 12.04.5 LTS and the compiler version is gcc > (Ubuntu/Linaro 4.6.4-1ubuntu1~12.04) 4.6.4. > > Abhishek > > > On Tue, Mar 17, 2015 at 5:06 PM, Jonathan Perkins < > perkinjo(a)cse.ohio-state.edu> wrote: > > > Thank you for your report. When trying to reproduce this problem > > locally I am running into a similar but different build error. Can > > you provide us more information on how you're building? Please let > > us know the Linux distribution and version and vendor of the > > compiler that you're using. > > > > On Tue, Mar 17, 2015 at 2:02 PM Abhishek Kulkarni > > <adkulkar(a)indiana.edu> wrote: > > > >> Hello, > >> > >> While trying to build mvapich 2.2rc2, I run into a build failure > >> when I use the following two flags together: --with-device=ch3:psm > >> --enable-thread-cs=per-object > >> > >> Is this a known issue? Since I expect to mostly run in a hybrid > >> (OpenMP+MPI) configuration, I am assuming that the granularity of > >> critical sections might be critical for good performance. > >> > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_acc_ops.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_ops.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_reqops.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_sync.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rndv.lo > >> src/mpid/ch3/src/ch3u_handle_recv_req.c: In function > >> ‘MPIDI_CH3_ReqHandler_FOPComplete’: > >> src/mpid/ch3/src/ch3u_handle_recv_req.c:963:9: warning: passing > >> argument 4 of ‘MPIDI_CH3_iStartMsgv’ from incompatible pointer > >> type [enabled by default] > >> ./src/mpid/ch3/include/mpidimpl.h:1580:5: note: expected ‘struct > >> MPID_Request **’ but argument is of type ‘struct MPID_Request *’ > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_eager.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_eagersync.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_win_fns.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-mpid_abort.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-mpid_cancel_recv.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-mpid_cancel_send.lo > >> src/mpid/ch3/src/ch3u_rndv.c: In function > >> ‘MPIDI_CH3_PktHandler_RndvReqToSend’: > >> src/mpid/ch3/src/ch3u_rndv.c:210:28: error: invalid operands to > >> binary == (have ‘MPID_cc_t’ and ‘int’) > >> make[2]: *** [src/mpid/ch3/src/lib_libmpi_la-ch3u_rndv.lo] Error 1 > >> make[2]: *** Waiting for unfinished jobs.... > >> src/mpid/ch3/src/ch3u_eager.c: In function > >> ‘MPIDI_CH3_PktHandler_EagerSend’: > >> src/mpid/ch3/src/ch3u_eager.c:880:28: error: invalid operands to > >> binary == (have ‘MPID_cc_t’ and ‘int’) > >> src/mpid/ch3/src/ch3u_eager.c: In function > >> ‘MPIDI_CH3_PktHandler_ReadySend’: > >> src/mpid/ch3/src/ch3u_eager.c:1022:28: error: invalid operands to > >> binary == (have ‘MPID_cc_t’ and ‘int’) > >> make[2]: *** [src/mpid/ch3/src/lib_libmpi_la-ch3u_eager.lo] Error 1 > >> src/mpid/ch3/src/ch3u_eagersync.c: In function > >> ‘MPIDI_CH3_PktHandler_EagerSyncSend’: > >> src/mpid/ch3/src/ch3u_eagersync.c:259:28: error: invalid operands > >> to binary == (have ‘MPID_cc_t’ and ‘int’) > >> make[2]: *** [src/mpid/ch3/src/lib_libmpi_la-ch3u_eagersync.lo] > >> Error 1 > >> > >> Thanks, > >> Abhishek > >> _______________________________________________ > >> mvapich-discuss mailing list > >> mvapich-discuss(a)cse.ohio-state.edu > >> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss > >> > > -- Jian Lin http://linjian.org _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mvapich-discuss] mvapich-2.2rc2 build failure with psm/per-object thread-cs.
by Jian Lin 26 Mar '15

26 Mar '15

Hi, Abhishek, Thanks for your note. By using the full configure options you provided, we are able to reproduce this behavior. However, MVAPICH2 2.1rc2 is based on MPICH 3.1.4. We found that the error also exists with the MPICH 3.1.4/3.2b1 even if using ch3:nemesis device. It will be good if the MPICH team resolves this. I am cc'ing this note to the MPICH team. Hi, MPICH community, The forwarded compilation error reported by the user is related to unmatched MPID_cc_t typedefs if some thread-related configure options are used. The default build without these thread-related options does not have any issue. Will you please take a look. Thanks! On Mon, 23 Mar 2015 20:29:12 -0400 Abhishek Kulkarni <adkulkar(a)indiana.edu> wrote: > Jonathan, > > I am using the following configure flags: > > ./configure --enable-threads=multiple --enable-async-progress > --with-device=ch3:psm --enable-thread-cs=per-object CFLAGS=-O3 > --enable-refcount=lock-free --enable-handle-allocation=tls > --with-atomic-primitives > > The distribution is Ubuntu 12.04.5 LTS and the compiler version is gcc > (Ubuntu/Linaro 4.6.4-1ubuntu1~12.04) 4.6.4. > > Abhishek > > > On Tue, Mar 17, 2015 at 5:06 PM, Jonathan Perkins < > perkinjo(a)cse.ohio-state.edu> wrote: > > > Thank you for your report. When trying to reproduce this problem > > locally I am running into a similar but different build error. Can > > you provide us more information on how you're building? Please let > > us know the Linux distribution and version and vendor of the > > compiler that you're using. > > > > On Tue, Mar 17, 2015 at 2:02 PM Abhishek Kulkarni > > <adkulkar(a)indiana.edu> wrote: > > > >> Hello, > >> > >> While trying to build mvapich 2.2rc2, I run into a build failure > >> when I use the following two flags together: --with-device=ch3:psm > >> --enable-thread-cs=per-object > >> > >> Is this a known issue? Since I expect to mostly run in a hybrid > >> (OpenMP+MPI) configuration, I am assuming that the granularity of > >> critical sections might be critical for good performance. > >> > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_acc_ops.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_ops.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_reqops.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rma_sync.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_rndv.lo > >> src/mpid/ch3/src/ch3u_handle_recv_req.c: In function > >> ‘MPIDI_CH3_ReqHandler_FOPComplete’: > >> src/mpid/ch3/src/ch3u_handle_recv_req.c:963:9: warning: passing > >> argument 4 of ‘MPIDI_CH3_iStartMsgv’ from incompatible pointer > >> type [enabled by default] > >> ./src/mpid/ch3/include/mpidimpl.h:1580:5: note: expected ‘struct > >> MPID_Request **’ but argument is of type ‘struct MPID_Request *’ > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_eager.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_eagersync.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-ch3u_win_fns.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-mpid_abort.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-mpid_cancel_recv.lo > >> CC src/mpid/ch3/src/lib_libmpi_la-mpid_cancel_send.lo > >> src/mpid/ch3/src/ch3u_rndv.c: In function > >> ‘MPIDI_CH3_PktHandler_RndvReqToSend’: > >> src/mpid/ch3/src/ch3u_rndv.c:210:28: error: invalid operands to > >> binary == (have ‘MPID_cc_t’ and ‘int’) > >> make[2]: *** [src/mpid/ch3/src/lib_libmpi_la-ch3u_rndv.lo] Error 1 > >> make[2]: *** Waiting for unfinished jobs.... > >> src/mpid/ch3/src/ch3u_eager.c: In function > >> ‘MPIDI_CH3_PktHandler_EagerSend’: > >> src/mpid/ch3/src/ch3u_eager.c:880:28: error: invalid operands to > >> binary == (have ‘MPID_cc_t’ and ‘int’) > >> src/mpid/ch3/src/ch3u_eager.c: In function > >> ‘MPIDI_CH3_PktHandler_ReadySend’: > >> src/mpid/ch3/src/ch3u_eager.c:1022:28: error: invalid operands to > >> binary == (have ‘MPID_cc_t’ and ‘int’) > >> make[2]: *** [src/mpid/ch3/src/lib_libmpi_la-ch3u_eager.lo] Error 1 > >> src/mpid/ch3/src/ch3u_eagersync.c: In function > >> ‘MPIDI_CH3_PktHandler_EagerSyncSend’: > >> src/mpid/ch3/src/ch3u_eagersync.c:259:28: error: invalid operands > >> to binary == (have ‘MPID_cc_t’ and ‘int’) > >> make[2]: *** [src/mpid/ch3/src/lib_libmpi_la-ch3u_eagersync.lo] > >> Error 1 > >> > >> Thanks, > >> Abhishek > >> _______________________________________________ > >> mvapich-discuss mailing list > >> mvapich-discuss(a)cse.ohio-state.edu > >> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss > >> > > -- Jian Lin http://linjian.org _______________________________________________ mvapich-discuss mailing list mvapich-discuss(a)cse.ohio-state.edu http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss

1 0

Problem with cpi example on multiple hosts
by HEINZ Josef (AREVA) 26 Mar '15

26 Mar '15

Hello everybody! First of all thank you very much for your endless answers. I have read and learned a lot about MPI. But I am discovering some problems with a MPI communication between two hosts. I am using the cpi example to exclude programming errors from my side. On both hosts mpich-3.0.4 is installed. I can access hostname2 from hostname1 through the ports xx:yy (in total 6 ports) using the command "ssh hostname2 -p <port>" The other way (from hostname2 to hostname1) I get an "connection refused" if I try "ssh hostname1 -p <port>" (same ports as before). But my admin assures me that the ports in the firewall are open for communication, only ssh is not listening. This should mpi do when starting up the tasks. Since the default port 22 does not work for ssh on hostname1, I configured my ~/.ssh/config (on hostname1) as followed to connect over port xx. Host hostname2 Hostname hostname2 PORT xx On hostname2 the ~/.ssh/config is empty since I can ssh through port 22. Before starting the process I restrict the port range as followed : export MPIEXEC_PORT_RANGE=xx:yy Starting the process: mpiexec -f hostfile -np 2 ./cpi Hostfile contains: Hostname1 Hostname2 This is the error output when running the example: Process 0 of 2 is on hostname1 Process 1 of 2 is on hostname2 Fatal error in PMPI_Reduce: A process has failed, error stack: PMPI_Reduce(1217)...............: MPI_Reduce(sbuf=0xbfca84f8, rbuf=0xbfca84f0, count=1, MPI_DOUBLE, MPI_SUM, root=0, MPI_COMM_WORLD) failed MPIR_Reduce_impl(1029)..........: MPIR_Reduce_intra(835)..........: MPIR_Reduce_binomial(144).......: MPIDI_CH3U_Recvq_FDU_or_AEP(667): Communication error with rank 1 Running cpi only on hostname1 or hostname2 is working without problems Process 0 of 2 is on hostname1 Process 1 of 2 is on hostname1 pi is approximately 3.1415926544231318, Error is 0.0000000008333387 wall clock time = 0.000408 I hope I provided enough information. Does anyone has an idea what setting is not correct? Best regards, Josef _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

SSH-free debugger launching with MPICH
by Daniel Everett 26 Mar '15

26 Mar '15

I am looking for a way to launch debuggers for all launched processes without using SSH. We debug the mpiexec starter process, so we can get MPIR_proctable back to determine which hosts the debuggers need to be on. I believe we could use mpiexec -f <hostfile> <debugger> to launch the debuggers. Initial investigations indicate that -launcher-exec does not do what we want. What is the best way to do this? Regards, Daniel Everett _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Re: [mpich-discuss] Problem linking an MPI program with an external shared library
by Isaila, Florin D. 23 Mar '15

23 Mar '15

With help from ALCF and using MPICH 3.1.4 I managed to solve the problem. Jeff, I needed dynamic libraries because I wanted to intercept some calls defined as weak symbols in other external libraries for tracing (e.g. malloc) and I did not find a form to accomplish this with static libraries. Now it works with dynamic libraries. Best regards Florin ________________________________________ From: Jeff Hammond [jeff.science(a)gmail.com] Sent: Thursday, March 19, 2015 10:40 AM To: discuss(a)mpich.org Subject: Re: [mpich-discuss] Problem linking an MPI program with an external shared library Do you need a dynamic library on BGQ? Why not just build static library? Best, Jeff On Thu, Mar 19, 2015 at 8:02 AM, Isaila, Florin D. <fisaila(a)mcs.anl.gov> wrote: > Hi, > > I have problems linking the an MPI program with an external shared library > in MPICH3.1 on Vesta BGQ. The error message I get is: > > mpicc -g -Wall -fPIC -Wextra -I. -c hello.c -o hello.o > mpicc -g -Wall -fPIC -Wextra -I. -c dyn_lib.c -o dyn_lib.o > mpicc -shared -Wl,-soname,libdyn_lib.so -o libdyn_lib.so dyn_lib.o > mpicc -g -Wall -fPIC -Wextra -I. -o hello hello.o -dynamic -Wl,-rpath, > -L. -ldyn_lib > /bgsys/drivers/toolchain/V1R2M2_base-efix014/gnu-linux/lib/gcc/powerpc64-bgq-linux/4.4.7/../../../../powerpc64-bgq-linux/bin/ld: > hello: hidden symbol `__dso_handle' in > /bgsys/drivers/toolchain/V1R2M2_base-efix014/gnu-linux/lib/gcc/powerpc64-bgq-linux/4.4.7/crtbegin.o > is referenced by DSO > /bgsys/drivers/toolchain/V1R2M2_base-efix014/gnu-linux/lib/gcc/powerpc64-bgq-linux/4.4.7/../../../../powerpc64-bgq-linux/bin/ld: > final link failed: Bad value > collect2: ld returned 1 exit status > make: *** [hello] Error 1 > > The problem seems to be caused by a __dso_handle symbol in the MPICH library > (both in static and dynamic libraries), but I do not know what the solution > is: > > [fisaila@vestalac2 lib]$ nm libmpich.so | grep __dso_handle > U __dso_handle > > > My MPICH3.1 is configured with: > ./configure > --prefix=/home/fisaila/software/mpich-3.1-THR_MULT-nodebug-install2rpath > --host=powerpc64-bgq-linux --with-device=pamid > --with-file-system=bg+bglockless --enable-thread-cs=per-object > --with-atomic-primitives --enable-handle-allocation=tls > --enable-refcount=lock-free --disable-predefined-refcount > CC=powerpc64-bgq-linux-gcc > > Any suggestion is welcome > > Thanks > Florin > > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Jeff Hammond jeff.science(a)gmail.com http://jeffhammond.github.io/ _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0