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

question about license of mpich and mpicxx
by ehsan saei 22 Oct '14

22 Oct '14

Dear all, I developed a code and parallelized it using mpi, I'd like to distribute my software but not the source code . I wanted to know whether MPICH is under GNU GPL license? (Then do I need to distribut my code under GPL license ). what about mpicxx that I use for making binary of my code? using mpicxx to compile the code forces the users to install MPICH on their machine doesn't it? Thanks in advance, Ehsan _______________________________________________ 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 Running MPI on cluster
by Md. Amjad Hossain 22 Oct '14

22 Oct '14

> Thank you so much Ken for your reply. > > Do I have to copy executable file to all machines? what I am doing is, > coding, compiling and running on a single machine and host_file contains > name of the other machines. > > I have run the command you gave me to run. it is printing the only name of > the machine I am executing command and showing all previous errors. Here > are the outputs after running the command twice: > > > [mhossain@md-lin-01 mpi_hello_world]$ /usr/lib64/mpich/bin/mpirun -n 4 -f > host_file hostname > md-lin-01.mcs.kent.edu > [proxy:0:[email protected]] launch_procs > (./pm/pmiserv/pmip_cb.c:648): unable to change wdir to > /home/mhossain/testMpi/mpi_hello_world (No such file or directory) > [proxy:0:[email protected]] HYD_pmcd_pmip_control_cmd_cb > (./pm/pmiserv/pmip_cb.c:893): launch_procs returned error > [proxy:0:[email protected]] HYDT_dmxu_poll_wait_for_event > (./tools/demux/demux_poll.c:77): callback returned error status > [proxy:0:[email protected]] main (./pm/pmiserv/pmip.c:206): demux > engine error waiting for event > [mpiexec(a)md-lin-01.mcs.kent.edu] control_cb > (./pm/pmiserv/pmiserv_cb.c:202): assert (!closed) failed > [mpiexec(a)md-lin-01.mcs.kent.edu] HYDT_dmxu_poll_wait_for_event > (./tools/demux/demux_poll.c:77): callback returned error status > [mpiexec(a)md-lin-01.mcs.kent.edu] HYD_pmci_wait_for_completion > (./pm/pmiserv/pmiserv_pmci.c:197): error waiting for event > [mpiexec(a)md-lin-01.mcs.kent.edu] main (./ui/mpich/mpiexec.c:331): process > manager error waiting for completion > > [mhossain@md-lin-01 mpi_hello_world]$ /usr/lib64/mpich/bin/mpirun -n 4 -f > host_file hostname > md-lin-01.mcs.kent.edu > [mpiexec(a)md-lin-01.mcs.kent.edu] control_cb > (./pm/pmiserv/pmiserv_cb.c:202): assert (!closed) failed > [mpiexec(a)md-lin-01.mcs.kent.edu] HYDT_dmxu_poll_wait_for_event > (./tools/demux/demux_poll.c:77): callback returned error status > [mpiexec(a)md-lin-01.mcs.kent.edu] HYD_pmci_wait_for_completion > (./pm/pmiserv/pmiserv_pmci.c:197): error waiting for event > [mpiexec(a)md-lin-01.mcs.kent.edu] main (./ui/mpich/mpiexec.c:331): process > manager error waiting for completion > > > Is this problem with configuration or MPICH version? > > Regards > Amjad > > > > Does your mpi_hello_world binary exist in the same directory on all the > machines you are trying to run on? Can you try running this: > > /usr/lib64/mpich/bin/mpirun -n 4 -f host_file hostname > > If it outputs the names of the hosts in your hostfile, we can be > confident that your mpirun and ssh setup is functioning correctly. > > Ken > > > On 10/21/2014 12:09 AM, Md. Amjad Hossain wrote: > > Hi I am trying to run simple hello world program on cluster nodes. I am > > running it by following command but getting errors: > > > > Command: /usr/lib64/mpich/bin/mpirun -n 4 -f host_file ./mpi_hello_world > > > > errors: > > [mpiexec(a)md-lin-01.mcs.kent.edu <mailto:[email protected]>] > > control_cb (./pm/pmiserv/pmiserv_cb.c:202): assert (!closed) failed > > [mpiexec(a)md-lin-01.mcs.kent.edu <mailto:[email protected]>] > > HYDT_dmxu_poll_wait_for_event (./tools/demux/demux_poll.c:77): callback > > returned error status > > [mpiexec(a)md-lin-01.mcs.kent.edu <mailto:[email protected]>] > > HYD_pmci_wait_for_completion (./pm/pmiserv/pmiserv_pmci.c:197): error > > waiting for event > > [mpiexec(a)md-lin-01.mcs.kent.edu <mailto:[email protected]>] > > main (./ui/mpich/mpiexec.c:331): process manager error waiting for > > completion > > > > > > Before running the command I am setting variables MPIRUN =mpi diretory > > and MPI_HOSTS=host_file. The "host_file" has four nodes and they can ssh > > to each other without password. > > > > MPICH version I am running is: 3.0.4. The MPI code is attached. > > > > Any help to solve the problem please? > > > > > > > > > > > > > _______________________________________________ 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 on make command to install mpich
by Sumit Sharma 22 Oct '14

22 Oct '14

Dear Sangmin, Thanks a ton it worked. Mpich installed... but when I am running WRF.exe now .. it is showing to use both the CPUs but the processing time remains same... dunno wats the reason. Best Regards Sumit Sharma Fellow & Area Convenor Centre for Environmental Studies Earth Science and Climate Change Division The Energy and Resources Institute (TERI) IHC complex, Lodi Road, New Delhi-3 Phone: 011-24682100, 24682111 Fax: 011-24682144 From: "Seo, Sangmin" <sseo(a)anl.gov> To: "<discuss(a)mpich.org>" <discuss(a)mpich.org> Date: 10/20/2014 08:52 PM Subject: Re: [mpich-discuss] Error on make command to install mpich Hi Sumit, In your c.txt, line 242-254 are: checking valgrind/drd.h usability... no checking valgrind/drd.h presence... no checking for valgrind/drd.h... no checking whether the valgrind headers are broken or too old... yes I think your system may have an old version of valgrind or some headers are broken. You may want to try configure with '--without-valgrind' if you don’t need the valgrind support, or you can try to build MPICH after installing a new version of valgrind. Best regards, -- Sangmin Seo Postdoctoral Appointee Mathematics and Computer Science Division Argonne National Laboratory http://www.mcs.anl.gov/~sseo/ On Oct 20, 2014, at 5:53 AM, Sumit Sharma <sumits(a)teri.res.in> wrote: Dear Researchers, I am trying to install mpich on my dual core linux machine. and getting errors when I try to run the make command. The c.txt and m.txt fiels are attached. I am using gcc compiler for the run. Looking forward to your kind help. Thanks (See attached file: m.txt)(See attached file: c.txt) Best Regards Sumit Sharma Fellow & Area Convenor Centre for Environmental Studies Earth Science and Climate Change Division The Energy and Resources Institute (TERI) IHC complex, Lodi Road, New Delhi-3 Email Security, formerly MX Logic.(See attached file: m.txt)(See attached file: c.txt) ------------------------------------------------------------------------------------------------------------ Disclaimer: The information contained in this e-mail is intended for the person or entity to which it is addressed, and it may contain confidential and/or privileged material. Any review or other use of this mail or taking any action based on it by persons or entities other than the intended recipient is strictly prohibited. If you receive this e-mail by mistake, please contact the sender, and delete all copies of this mail.This e-mail has been scanned and verified by McAfee SaaS Email Security, formerly MX Logic.<m.txt><c.txt>_______________________________________________ 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 ------------------------------------------------------------------------------------------------------------ Disclaimer: The information contained in this e-mail is intended for the person or entity to which it is addressed, and it may contain confidential and/or privileged material. Any review or other use of this mail or taking any action based on it by persons or entities other than the intended recipient is strictly prohibited. If you receive this e-mail by mistake, please contact the sender, and delete all copies of this mail.This e-mail has been scanned and verified by McAfee SaaS Email Security, formerly MX Logic. _______________________________________________ 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] MPICH Examples
by Wesley Bland 21 Oct '14

21 Oct '14

It depends on which tests / examples you’re talking about. The codes in test/ are generally for regression tests for our own development. They have descriptions in each of the files in the comments to explain what they do, but there is no universal documentation for the entire directory. For the codes in examples/, there is a README file that gives a quick description of what each test tries to accomplish. Beyond that, there might be some comments in the codes themselves. Thanks, Wesley > On Oct 21, 2014, at 10:45 AM, Michael L. Stokes <Michael.Stokes(a)uah.edu> wrote: > > Is there any description of what each test accomplishes? > > Thanks > Mike > _______________________________________________ > 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 Examples
by Michael L. Stokes 21 Oct '14

21 Oct '14

Is there any description of what each test accomplishes? Thanks Mike _______________________________________________ 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 Running MPI on cluster
by Kenneth Raffenetti 21 Oct '14

21 Oct '14

Does your mpi_hello_world binary exist in the same directory on all the machines you are trying to run on? Can you try running this: /usr/lib64/mpich/bin/mpirun -n 4 -f host_file hostname If it outputs the names of the hosts in your hostfile, we can be confident that your mpirun and ssh setup is functioning correctly. Ken On 10/21/2014 12:09 AM, Md. Amjad Hossain wrote: > Hi I am trying to run simple hello world program on cluster nodes. I am > running it by following command but getting errors: > > Command: /usr/lib64/mpich/bin/mpirun -n 4 -f host_file ./mpi_hello_world > > errors: > [mpiexec(a)md-lin-01.mcs.kent.edu <mailto:[email protected]>] > control_cb (./pm/pmiserv/pmiserv_cb.c:202): assert (!closed) failed > [mpiexec(a)md-lin-01.mcs.kent.edu <mailto:[email protected]>] > HYDT_dmxu_poll_wait_for_event (./tools/demux/demux_poll.c:77): callback > returned error status > [mpiexec(a)md-lin-01.mcs.kent.edu <mailto:[email protected]>] > HYD_pmci_wait_for_completion (./pm/pmiserv/pmiserv_pmci.c:197): error > waiting for event > [mpiexec(a)md-lin-01.mcs.kent.edu <mailto:[email protected]>] > main (./ui/mpich/mpiexec.c:331): process manager error waiting for > completion > > > Before running the command I am setting variables MPIRUN =mpi diretory > and MPI_HOSTS=host_file. The "host_file" has four nodes and they can ssh > to each other without password. > > MPICH version I am running is: 3.0.4. The MPI code is attached. > > Any help to solve the problem please? > > > > > > > > _______________________________________________ > 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] Re : MPI process killed and SIGUSR1
by Wesley Bland 21 Oct '14

21 Oct '14

That should also return an error, but as I said previously, the call to wait shouldn't be blocking either. > On Oct 20, 2014, at 11:09 PM, Roy, Hirak <Hirak_Roy(a)mentor.com> wrote: > > Will MPI_IProbe going to work here? I don’t want to block my thread on a receive. The client can be busy arbitrary amount of time and the server-thread can not be blocked on a receive. > > Thanks, > Hirak > > From: Roy, Hirak > Sent: Tuesday, October 21, 2014 9:37 AM > To: discuss(a)mpich.org > Subject: Re : MPI process killed and SIGUSR1 > > Hi Wesley, > > So if I do the following from the server side : > > MPI_Irecv (&buf, 1, MPI_INT, 0, MPI_TAG_SANITY, clientCom, &req); > MPI_Wait(&req, &stat); > > > > Now the client is killed. However, the wait does not terminate. > > Thanks, > Hirak > > > > It's not guaranteed that you will receive an error from a send (of any kind other than ssend). The reason is that the implementation can buffer the message internally and return to the application before the message is actually transferred. You will definitely receive an error related to any kind of receive, but sends are less reliable in that sense. > > Thanks, > Wesley > > > > On Oct 20, 2014, at 7:53 AM, Roy, Hirak <Hirak_Roy at mentor.com<mailto:Hirak_Roy at mentor.com>> wrote: > > > Hi Wesley, > > > > Suppose my server-client connection is setup using MPI_Comm_accept/MPI_Comm_connect. > > Now one of the clients dies. > > If the server sends a non-blocking message to the dead client by MPI_Isend, will MPI_Wait guaranteed to return Error code? > > > > Thanks, > > Hirak > > > > > > > > > > >You will only receive the notification if you use a communicating call. That call doesn't do any communication so it won't return the error. > > > _______________________________________________ > 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] Re : MPI process killed and SIGUSR1
by Wesley Bland 21 Oct '14

21 Oct '14

So you're killing rank 0 while the other rank is blocked in the MPI wait? If that's true, then it's a bug. We're working on some other problems with the ft implementation right now. We'll take a look at that as well. > On Oct 20, 2014, at 11:07 PM, Roy, Hirak <Hirak_Roy(a)mentor.com> wrote: > > Hi Wesley, > > So if I do the following from the server side : > > MPI_Irecv (&buf, 1, MPI_INT, 0, MPI_TAG_SANITY, clientCom, &req); > MPI_Wait(&req, &stat); > > > > Now the client is killed. However, the wait does not terminate. > > Thanks, > Hirak > > > > It's not guaranteed that you will receive an error from a send (of any kind other than ssend). The reason is that the implementation can buffer the message internally and return to the application before the message is actually transferred. You will definitely receive an error related to any kind of receive, but sends are less reliable in that sense. > > Thanks, > Wesley > > > > On Oct 20, 2014, at 7:53 AM, Roy, Hirak <Hirak_Roy at mentor.com<mailto:Hirak_Roy at mentor.com>> wrote: > > > Hi Wesley, > > > > Suppose my server-client connection is setup using MPI_Comm_accept/MPI_Comm_connect. > > Now one of the clients dies. > > If the server sends a non-blocking message to the dead client by MPI_Isend, will MPI_Wait guaranteed to return Error code? > > > > Thanks, > > Hirak > > > > > > > > > > >You will only receive the notification if you use a communicating call. That call doesn't do any communication so it won't return the error. > > > _______________________________________________ > 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] In-Reply-To: 0A407957589BAB4F924824150C4293EF4C497F0F@CITESMBX3.ad.uillinois.edu
by Wesley Bland 21 Oct '14

21 Oct '14

Initially, yes it will only be available in MPICH proper. However, eventually all of the derivatives will pick up that code as well. Thanks, Wesley > On Oct 21, 2014, at 3:23 AM, Victor Vysotskiy <victor.vysotskiy(a)teokem.lu.se> wrote: > > Hi Xin, > > >I think it is due to a bug for MPI_PUT to handle large derived datatype and we will fix it as soon as possible. > > Thanks! Apparently, almost all MPICH derivatives are suffering from this bug. In particular, the problem can be reproduced within the Open ParaStation MPI, MVAPICH, and Intel MPI libraries. Do I understand correctly that the fix will be available only for the MPICH3 branch? > > With best regards, > Victor. > _______________________________________________ > 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

In-Reply-To: 0A407957589BAB4F924824150C4293EF4C497F0F@CITESMBX3.ad.uillinois.edu
by Victor Vysotskiy 21 Oct '14

21 Oct '14

Hi Xin, >I think it is due to a bug for MPI_PUT to handle large derived datatype and we will fix it as soon as possible. Thanks! Apparently, almost all MPICH derivatives are suffering from this bug. In particular, the problem can be reproduced within the Open ParaStation MPI, MVAPICH, and Intel MPI libraries. Do I understand correctly that the fix will be available only for the MPICH3 branch? With best regards, Victor. _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0