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

Re: [mpich-discuss] MPI_DIMS_CREATE
by Andreas Schäfer 31 Mar '15

31 Mar '15

Good catch. If I'm not mistaken then the line n_root = 1 << (n_root / 2); in src/mpi/topo/dims_create.c sets an overly strict boundary as the division by 2 will round down. In the test case n_root will be set to 16, which is smaller than 19. The following line should fix this, while being only marginally slower (with prime number density being low for large n...) n_root = 1 << (n_root / 2 + 1); Cheers -Andreas On 09:10 Tue 31 Mar , Valery wrote: > Dear All > > I noticed that MPI_DIMS_CREATE cannot split eg > 361 > into a 19 * 19 processor grid. Is that a feature? > > The code follows. > > I used > MPICH-3.1.4 > gcc-4.9.2 > > valery > > > cat mpi_dims.f90 > program test > use mpi > implicit none > integer :: ierr, nnodes, ndims, dims(2), i > call MPI_INIT( ierr ) > ndims = 2 > do i = 1, 200 > nnodes = i**2 > dims(:) = 0 > call MPI_DIMS_CREATE( nnodes, ndims, dims, ierr ) > if( dims(1) /= i ) write(*,*) i, dims > enddo > call MPI_FINALIZE(ierr) > end program test > > mpif90 mpi_dims.f90 > > > mpiexec -n 1 ./a.out > 19 361 1 > 41 1681 1 > 43 1849 1 > 71 5041 1 > 73 5329 1 > 79 6241 1 > 83 6889 1 > 89 7921 1 > 137 18769 1 > 139 19321 1 > 149 22201 1 > 151 22801 1 > 157 24649 1 > 163 26569 1 > 167 27889 1 > 173 29929 1 > 179 32041 1 > 181 32761 1 > > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- ========================================================== Andreas Schäfer HPC and Grid Computing Chair of Computer Science 3 Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany +49 9131 85-27910 PGP/GPG key via keyserver http://www.libgeodecomp.org ========================================================== (\___/) (+'.'+) (")_(") This is Bunny. Copy and paste Bunny into your signature to help him gain world domination! _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

MPI_DIMS_CREATE
by Valery 31 Mar '15

31 Mar '15

Dear All I noticed that MPI_DIMS_CREATE cannot split eg 361 into a 19 * 19 processor grid. Is that a feature? The code follows. I used MPICH-3.1.4 gcc-4.9.2 valery cat mpi_dims.f90 program test use mpi implicit none integer :: ierr, nnodes, ndims, dims(2), i call MPI_INIT( ierr ) ndims = 2 do i = 1, 200 nnodes = i**2 dims(:) = 0 call MPI_DIMS_CREATE( nnodes, ndims, dims, ierr ) if( dims(1) /= i ) write(*,*) i, dims enddo call MPI_FINALIZE(ierr) end program test mpif90 mpi_dims.f90 mpiexec -n 1 ./a.out 19 361 1 41 1681 1 43 1849 1 71 5041 1 73 5329 1 79 6241 1 83 6889 1 89 7921 1 137 18769 1 139 19321 1 149 22201 1 151 22801 1 157 24649 1 163 26569 1 167 27889 1 173 29929 1 179 32041 1 181 32761 1 _______________________________________________ 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] error: ‘MPIDI_CH3_ReqHandler_PutRecvComplete’ undeclared
by Zhao, Xin 30 Mar '15

30 Mar '15

Hi Christoph, IB netmod is contributed by RIKEN and they are working on cleaning up this netmod. At the meantime, I would recommend using MXM netmod instead by specifying --with-device=ch3:nemesis:mxm. MXM is a netmod for IB networks using the MXM API instead of Verbs to interact with the HCA. Note that it is only for Mellanox hardware. Thanks, Xin ________________________________________ From: Christoph Niethammer [niethammer(a)hlrs.de] Sent: Monday, March 30, 2015 6:30 AM To: discuss(a)mpich.org Subject: [mpich-discuss] error: ‘MPIDI_CH3_ReqHandler_PutRecvComplete’ undeclared Hello, I just tried to install MPICH 3.1.4 on an Infiniband system with ./configure --prefix=$INSTALLPATH --enable-fast=all,O3 --enable-shared --enable-static --with-pm=hydra --with-device=ch3:nemesis:ib When running make I get compilation errors in the ib files like the following: /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c: In function ‘MPID_nem_ib_PktHandler_Put’: /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:1839:24: error: ‘MPIDI_CH3_ReqHandler_PutRecvComplete’ undeclared (first use in this function) req->dev.OnFinal = MPIDI_CH3_ReqHandler_PutRecvComplete; /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:1839:24: note: each undeclared identifier is reported only once for each function it appears in /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:1847:20: error: ‘MPIDI_CH3_Pkt_put_t’ has no member named ‘immed_len’ if (put_pkt->immed_len > 0) { ^ In file included from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/include/mpid_nem_pre.h:12:0, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/include/mpidi_ch3_pre.h:9, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/include/mpidpre.h:49, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/include/mpiimpl.h:172, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/include/mpidimpl.h:22, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/include/mpidi_ch3_impl.h:10, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/include/mpid_nem_impl.h:11, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_impl.h:11, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:8: /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:1849:51: error: ‘MPIDI_CH3_Pkt_put_t’ has no member named ‘data’ MPIU_Memcpy(req->dev.user_buf, put_pkt->data, put_pkt->immed_len); ... Ist this problem already known? I could not find anything related to this except a renaming of functions which took place in Oct 2014: http://git.mpich.org/mpich.git/commitdiff/fe15ea266ef2e8a2d53565b88ab3ce4d4… Regards Christoph Niethammer _______________________________________________ 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

error: ‘MPIDI_CH3_ReqHandler_PutRecvComplete’ undeclared
by Christoph Niethammer 30 Mar '15

30 Mar '15

Hello, I just tried to install MPICH 3.1.4 on an Infiniband system with ./configure --prefix=$INSTALLPATH --enable-fast=all,O3 --enable-shared --enable-static --with-pm=hydra --with-device=ch3:nemesis:ib When running make I get compilation errors in the ib files like the following: /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c: In function ‘MPID_nem_ib_PktHandler_Put’: /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:1839:24: error: ‘MPIDI_CH3_ReqHandler_PutRecvComplete’ undeclared (first use in this function) req->dev.OnFinal = MPIDI_CH3_ReqHandler_PutRecvComplete; /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:1839:24: note: each undeclared identifier is reported only once for each function it appears in /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:1847:20: error: ‘MPIDI_CH3_Pkt_put_t’ has no member named ‘immed_len’ if (put_pkt->immed_len > 0) { ^ In file included from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/include/mpid_nem_pre.h:12:0, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/include/mpidi_ch3_pre.h:9, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/include/mpidpre.h:49, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/include/mpiimpl.h:172, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/include/mpidimpl.h:22, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/include/mpidi_ch3_impl.h:10, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/include/mpid_nem_impl.h:11, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_impl.h:11, from /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:8: /work/mpi/mpich/3.1.4-gnu-4.9.1/mpich-3.1.4/src/mpid/ch3/channels/nemesis/netmod/ib/ib_poll.c:1849:51: error: ‘MPIDI_CH3_Pkt_put_t’ has no member named ‘data’ MPIU_Memcpy(req->dev.user_buf, put_pkt->data, put_pkt->immed_len); ... Ist this problem already known? I could not find anything related to this except a renaming of functions which took place in Oct 2014: http://git.mpich.org/mpich.git/commitdiff/fe15ea266ef2e8a2d53565b88ab3ce4d4… Regards Christoph Niethammer _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

EuroMPI 2015 Call for Papers
by Brice Goglin 29 Mar '15

29 Mar '15

*********************************************** * EuroMPI 2015 Call for Papers * * The 22nd European MPI Users' Group Meeting * * Bordeaux, France * * 21st - 24th September, 2015 * * https://eurompi2015.bordeaux.inria.fr * *********************************************** *********************************************** * in-cooperation status with ACM and SIGHPC * *********************************************** BACKGROUND AND TOPICS --------------------- EuroMPI is the preeminent meeting for users, developers and researchers to interact and discuss new developments and applications of message-passing parallel computing, in particular in and related to the Message Passing Interface (MPI). The annual meeting has a long, rich tradition, and has been held in European countries. Following a 21st edition that took place in Japan, the 22nd meeting edition shall be back on European soil, in Bordeaux, France. EuroMPI 2015 will continue to focus on not just MPI, but also extensions or alternative interfaces for high-performance homogeneous/heterogeneous /hybrid systems, benchmarks, tools, parallel I/O, fault tolerance, and parallel applications using MPI and other interfaces. Through the presentation of contributed papers, poster presentations and invited talks, attendees will have the opportunity to share ideas and experiences to contribute to the improvement and furthering of message-passing and related parallel programming paradigms. In addition to the main conference's technical program, one-day or half-day workshops will be held. The Call for Workshops is announced separately and also shown in the conference page. Topics of interest for the meeting include, but are not limited to: - MPI implementation issues and improvements towards exascale computing, such as manycores, GPGPU, and heterogeneous architectures. - Extensions to and shortcomings of MPI. - Hybrid and heterogeneous programming with MPI and other interfaces. - Interaction between message-passing software and hardware, in particular new high performance architectures. - MPI support for data-intensive parallel applications. - New MPI-IO mechanisms and I/O stack optimizations. - Fault tolerance in message-passing implementations and systems. - Performance evaluation for MPI and MPI based applications. - Automatic performance tuning of MPI applications and implementations. - Verification of message passing applications and protocols. - Applications using message-passing, in particular in Computational Science and Scientific Computing. - Parallel algorithms in the message-passing paradigm. - New programming paradigms implemented over MPI, like hierarchical programming and global address spaces - MPI parallel programming in clouds - MPI applications performance on clouds IMPORTANT DATES --------------- - Workshop proposals due: April 10th, 2015 - Workshop notification: April 30th, 2015 - Submission of full papers and poster abstracts: May 4th, 2015 - Author notification: June 12th, 2015 - Camera Ready papers due: June 29th, 2015 - Tutorials: September 21st, 2015 - Conference: September 22nd-24th, 2015 SUBMISSION INSTRUCTIONS ----------------------- Contributors are invited to submit a full paper as a PDF document not exceeding 10 pages in English. The title page should contain an abstract of at most 200 words and five specific, topical keywords. The paper must be formatted according to double-column ACM ICPS proceedings style. The usage of LaTeX for preparation of the contribution as well as the submission in camera ready format is strongly recommended. Style files can be found at http://www.acm.org/publications/icps-instructions/. New work that is not yet mature for a full paper are invited for the poster session. Contributions to the poster session should be submitted in the form of a two page abstract. All contributions will be fully peer reviewed by the program committee. For submission, see https://www.easychair.org/conferences/?conf=eurompi2015. SPECIAL ISSUE ------------- The best papers of the workshop will be considered for inclusion in a special issue of International Journal of High Performance Computing Applications (IJHPCA). COMMITTEE --------- General chair: Jack Dongarra, University of Tennessee Program co-chairs: Alexandre Denis, Inria Bordeaux Sud-Ouest Brice Goglin, Inria Bordeaux Sud-Ouest Emmanuel Jeannot, Inria Bordeaux Sud-Ouest Guillaume Mercier, Bordeaux INP Program Committee: Ahmad Afsahi, Queen's University Pavan Balaji, Argonne National Laboratory Siegfried Benkner, University of Vienna George Bosilca, Innovative Computing Laboratory - University of Tennessee Ron Brightwell, Sandia National Laboratories Franck Cappello, Argonne National Laboratory and University of Illinois at Urbana Champaign Jesus Carretero, Universidad Carlos III de Madrid Yiannis Cotronis, National and Kapodistrian University of Athens Anthony Danalis, University of Tennessee Bronis de Supinski, Lawrence Livermore National Laboratory Alexandre Denis, Inria Bordeaux Sud Ouest David E. Singh, Universidad Carlos III de Madrid Christian Engelmann, Oak Ridge National Laboratory Edgar Gabriel, University of Houston Javier Garcia Blas, Carlos III University Stéphane Génaud, ENSIIE-Université de Strasbourg Balazs Gerofi, RIKEN AICS Brice Goglin, Inria Bordeaux Sud-Ouest David Goodell, Cisco Systems, Inc. Ganesh Gopalakrishnan, University of Utah Richard Graham, Mellanox Technologies William Gropp, University of Illinois at Urbana-Champaign Thomas Hérault, University of Tennessee Torsten Hoefler, ETH Zurich Atsushi Hori, RIKEN AICS Florin Isaila, Argonne National Laboratory and University Carlos III of Madrid Yutaka Ishikawa, RIKEN AICS Emmanuel Jeannot, Inria Bordeaux Sud-Ouest Rainer Keller, HFT Stuttgart Dries Kimpe, Argonne National Laboratory Julian Kunkel, Universität Hamburg Alexey Lastovetsky, University College Dublin Laurent Lefèvre, Inria Thomas Ludwig, University of Hamburg Ewing Lusk, Argonne National Laboratory Guillaume Mercier, INPB Bernd Mohr, Juelich Supercomputing Centre Julián Morillo, BSC Matthias Mueller, RWTH Aachen University Kengo Nakajima, The University of Tokyo Maria S. Perez, Universidad Politecnica de Madrid Rolf Rabenseifner, HLRS, University of Stuttgart Rolf Riesen, IBM Luis Miguel Sanchez, Universidad Carlos III de Madrid Mitsuhisa Sato, RIKEN AICS and University of Tsukuba Martin Schulz, Lawrence Livermore National Laboratory Anna Sikora, Universitat Autònoma de Barcelona Jeff Squyres, Cisco Systems, Inc. Shinji Sumimoto, Fujitsu Laboratories Frédéric Suter, CC IN2P3 / CNRS Guillermo Taboada, Torusware and Corunna University Masamichi Takagi, RIKEN AICS Domenico Talia, University of Calabria Rajeev Thakur, Argonne National Laboratory Jesper Larsson Träff, Vienna University of Technology (TU Wien) Denis Trystram, Grenoble university Yuichi Tsujita, RIKEN AICS Keith Underwood, Intel Alan Wagner, University of British Columbia Xin Yuan, Florida State University _______________________________________________ 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 Vaibhav Rekhate 28 Mar '15

28 Mar '15

I'll try to capture a photo next time. As screenshot won't work, I'll have to grab the image with a camera. Regards, Vaibhav Rekhate College of Engineering, Pune On Fri, Mar 27, 2015 at 9:10 AM -0700, "Kenneth Raffenetti" <raffenet(a)mcs.anl.gov<mailto:[email protected]>> wrote: Are you able to capture the debug or panic output from the nodes before they reboot? It is difficult to diagnose the issue without that information. On 03/27/2015 10:52 AM, Vaibhav Rekhate wrote: > I have setup a cluster on Ubuntu 12.04 machines (using the steps > outlined here: https://help.ubuntu.com/community/MpichCluster) > > The environment variable PATH is set. The code is compiled on the master > node and executed on the master node. After execution of the code, the > slave nodes crash, i.e. the slaves start rebooting, debug statements on > the display indicate kernel panic. The slaves have to be restarted. This > is not exactly reproducible as it does not happen every time, but sometimes. > > > Regards, > Vaibhav Rekhate > B. Tech, Computer Engineering > College of Engineering, Pune > ------------------------------------------------------------------------ > *From:* Huiwei Lu <huiweilu(a)mcs.anl.gov> > *Sent:* 27 March 2015 07:41 PM > *To:* discuss(a)mpich.org > *Subject:* Re: [mpich-discuss] Slave hosts panic after mpirun > 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 <mailto:[email protected]>> 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 <mailto:[email protected]> > 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 > _______________________________________________ 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 Huiwei Lu 27 Mar '15

27 Mar '15

Hi Jian and Abhishek, The build problem should have been fixed in the latest mpich master branch. You can find the change here: http://trac.mpich.org/projects/mpich/changeset/983a60e8fd6490daf15301f5f092… Please let me know if it works. Thanks, -- Huiwei Lu Postdoc Appointee Mathematics and Computer Science Division Argonne National Laboratory On Thu, Mar 26, 2015 at 7:56 PM, Huiwei Lu <huiweilu(a)mcs.anl.gov> wrote: > 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

un-subscribe
by luoyandaer 27 Mar '15

27 Mar '15

un-subscribe _______________________________________________ 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 Kenneth Raffenetti 27 Mar '15

27 Mar '15

Are you able to capture the debug or panic output from the nodes before they reboot? It is difficult to diagnose the issue without that information. On 03/27/2015 10:52 AM, Vaibhav Rekhate wrote: > I have setup a cluster on Ubuntu 12.04 machines (using the steps > outlined here: https://help.ubuntu.com/community/MpichCluster) > > The environment variable PATH is set. The code is compiled on the master > node and executed on the master node. After execution of the code, the > slave nodes crash, i.e. the slaves start rebooting, debug statements on > the display indicate kernel panic. The slaves have to be restarted. This > is not exactly reproducible as it does not happen every time, but sometimes. > > > Regards, > Vaibhav Rekhate > B. Tech, Computer Engineering > College of Engineering, Pune > ------------------------------------------------------------------------ > *From:* Huiwei Lu <huiweilu(a)mcs.anl.gov> > *Sent:* 27 March 2015 07:41 PM > *To:* discuss(a)mpich.org > *Subject:* Re: [mpich-discuss] Slave hosts panic after mpirun > 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 <mailto:[email protected]>> 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 <mailto:[email protected]> > 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 > _______________________________________________ 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 Vaibhav Rekhate 27 Mar '15

27 Mar '15

I have setup a cluster on Ubuntu 12.04 machines (using the steps outlined here: https://help.ubuntu.com/community/MpichCluster) The environment variable PATH is set. The code is compiled on the master node and executed on the master node. After execution of the code, the slave nodes crash, i.e. the slaves start rebooting, debug statements on the display indicate kernel panic. The slaves have to be restarted. This is not exactly reproducible as it does not happen every time, but sometimes. Regards, Vaibhav Rekhate B. Tech, Computer Engineering College of Engineering, Pune ________________________________ From: Huiwei Lu <huiweilu(a)mcs.anl.gov> Sent: 27 March 2015 07:41 PM To: discuss(a)mpich.org Subject: Re: [mpich-discuss] Slave hosts panic after mpirun 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<mailto:[email protected]>> 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<mailto:[email protected]> 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