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

Re: [mpich-discuss] pgf90-Error-Unknown switch: -pthread
by Gus Correa 09 Jul '14

09 Jul '14

Hi James Adding '-noswitcherror' to the CFLAGS, FFLAGS, FCFLAGS, may be a workaround to this problem. My two cents, Gus Correa On 07/09/2014 05:55 PM, James Long wrote: > I need to use srun to launch mpi jobs under slurm, so configured mpich-3.1.1 with > > $ env CC=pgcc FC=pgf90 CXX=pgCC CPPFLAGS="-DNDEBUG -DpgiFortran" CFLAGS="-O2" FFLAGS="-O2 -w" ./configure --prefix=/opt/mpich-slurm --with-pmi=slurm --with-pm=no --with-slurm-include=/usr/include/slurm --with-slurm-lib=/usr/lib/slurm --enable-fortran=yes > > The following error occurs when compiling: > > CC src/mpid/common/datatype/lib_libmpi_la-mpir_type_flatten.lo > CC src/mpid/common/sched/lib_libmpi_la-mpid_sched.lo > CC src/mpid/common/thread/lib_libmpi_la-mpid_thread.lo > CC src/mpi_t/lib_libmpi_la-mpit.lo > CC src/nameserv/pmi/lib_libmpi_la-pmi_nameserv.lo > GEN lib/libmpi.la > pgf90-Error-Unknown switch: -pthread > make[2]: *** [lib/libmpi.la] Error 1 > make[2]: Leaving directory `/home/boss/mpich-3.1.1' > make[1]: *** [all-recursive] Error 1 > make[1]: Leaving directory `/home/boss/mpich-3.1.1' > make: *** [all] Error 2 > > config.log is at http://pastebin.com/aB8CZsr0 > > Jim > -- > > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% > James Long > International Arctic Research Center > University of Alaska Fairbanks > jlong|at|alaska.edu > (907) 474-2440 > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% > > > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss >

1 0

pgf90-Error-Unknown switch: -pthread
by James Long 09 Jul '14

09 Jul '14

I need to use srun to launch mpi jobs under slurm, so configured mpich-3.1.1 with $ env CC=pgcc FC=pgf90 CXX=pgCC CPPFLAGS="-DNDEBUG -DpgiFortran" CFLAGS="-O2" FFLAGS="-O2 -w" ./configure --prefix=/opt/mpich-slurm --with-pmi=slurm --with-pm=no --with-slurm-include=/usr/include/slurm --with-slurm-lib=/usr/lib/slurm --enable-fortran=yes The following error occurs when compiling: CC src/mpid/common/datatype/lib_libmpi_la-mpir_type_flatten.lo CC src/mpid/common/sched/lib_libmpi_la-mpid_sched.lo CC src/mpid/common/thread/lib_libmpi_la-mpid_thread.lo CC src/mpi_t/lib_libmpi_la-mpit.lo CC src/nameserv/pmi/lib_libmpi_la-pmi_nameserv.lo GEN lib/libmpi.la pgf90-Error-Unknown switch: -pthread make[2]: *** [lib/libmpi.la] Error 1 make[2]: Leaving directory `/home/boss/mpich-3.1.1' make[1]: *** [all-recursive] Error 1 make[1]: Leaving directory `/home/boss/mpich-3.1.1' make: *** [all] Error 2 config.log is at http://pastebin.com/aB8CZsr0 Jim -- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% James Long International Arctic Research Center University of Alaska Fairbanks jlong|at|alaska.edu (907) 474-2440 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1 0

Re: [mpich-discuss] Does mpich support zero-copy data transfers over ethernet (packet mmap)?
by "Antonio J. Peña" 09 Jul '14

09 Jul '14

Hi Clemens, Our TCP-based netmods don't have this feature of packet mmap. Are you sure that the high CPU load you are seeing is not caused by active polling for completion? Otherwise, could you share with us your profiling results? Antonio On 07/09/2014 07:57 AM, Clemens Eisserer wrote: > Hi, > > We've recently conducted some tests showing that CPU overhead is > significant when sending/receiving data via TCP or UDP over 10GigE > cards on Linux (20-50% of an Ivy Bridge core clocked at 3Ghz). > A lot of time is spent copying data from kernel to user space and vice versa. > > As we would like to avoid this overhead as far as possible, does > mpich2 support zero copy data transfer using packet mmap? > > Thank you in advance, Clemens > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Antonio J. Peï¿½a Postdoctoral Appointee Mathematics and Computer Science Division Argonne National Laboratory 9700 South Cass Avenue, Bldg. 240, Of. 3148 Argonne, IL 60439-4847 apenya(a)mcs.anl.gov www.mcs.anl.gov/~apenya

1 0

Does mpich support zero-copy data transfers over ethernet (packet mmap)?
by Clemens Eisserer 09 Jul '14

09 Jul '14

Hi, We've recently conducted some tests showing that CPU overhead is significant when sending/receiving data via TCP or UDP over 10GigE cards on Linux (20-50% of an Ivy Bridge core clocked at 3Ghz). A lot of time is spent copying data from kernel to user space and vice versa. As we would like to avoid this overhead as far as possible, does mpich2 support zero copy data transfer using packet mmap? Thank you in advance, Clemens

1 0

unknown type name for mpich-master-v3.1.1-71-gc244ba49
by Siegmar Gross 09 Jul '14

09 Jul '14

Hi, today I tried to install mpich-master-v3.1.1-71-gc244ba49 on my Linux machine (openSUSE 12.1 x86_64). "make" fails with the following error for gcc-4.9.0. ... CC src/mpi/attr/lib_libmpi_la-attr_delete.lo In file included from ../mpich-master-v3.1.1-71-gc244ba49/src/mpi/attr/attr_delete.c:8:0: ../mpich-master-v3.1.1-71-gc244ba49/src/include/mpiimpl.h:105:12: error: unknown type name 'useconds_t' int usleep(useconds_t usec); ^ make[2]: *** [src/mpi/attr/lib_libmpi_la-attr_delete.lo] Error 1 ... I used the following configure-command. ../mpich-master-v3.1.1-71-gc244ba49/configure \ --prefix=/usr/local/mpich-3.1.2_64_gcc \ --libdir=/usr/local/mpich-3.1.2_64_gcc/lib64 \ --includedir=/usr/local/mpich-3.1.2_64_gcc/include64 \ CC="gcc" CXX="g++" F77="gfortran" FC="gfortran" \ CFLAGS="-m64 -std=c11" CXXFLAGS="-m64" FFLAGS="-m64" FCFLAGS="-m64" \ LDFLAGS="-m64" \ --enable-f77 --enable-fc --enable-cxx --enable-romio \ --enable-debuginfo --enable-smpcoll \ --enable-threads=runtime --with-thread-package=posix \ --enable-shared \ |& tee log.configure.$SYSTEM_ENV.$MACHINE_ENV.64_gcc The log-file contains no errors, but some warnings. linpc1 mpich-3.1.2-Linux.x86_64.64_gcc 118 grep -i warning \ log.configure.Linux.x86_64.64_gcc configure: WARNING: Some versions of gcc do not include debugging information configure: WARNING: Structures containing long doubles may be aligned differently from structures with floats or longs. MPICH does not handle this case automatically and you should avoid assumed extents for structures containing float types. configure: WARNING: File locks may not work with NFS. See the Installation and configure: WARNING: "-fvisibility=hidden" has been added to the hwloc CFLAGS checking warnings... no configure: WARNING: The ch3 device did not set the maximum size of an error string, 512 being used. config.status: WARNING: '../mpich-master-v3.1.1-71-gc244ba49/src/packaging/envmods/mpich.module.in' seems to ignore the --datarootdir setting linpc1 mpich-3.1.2-Linux.x86_64.64_gcc 119 I would be grateful if somebody could fix the problem. Thank you very much for your help in advance. Kind regards Siegmar

1 0

Re: [mpich-discuss] Problem with "sock"-channel
by Thakur, Rajeev 08 Jul '14

08 Jul '14

If you are running on a single machine, Nemesis will use shared memory, so no TCP connection issues. ch3:sock will always use TCP and try to set up a socket connection. Rajeev On Jul 8, 2014, at 9:16 AM, Klaus Philip <Philip.Klaus.fl(a)ait.ac.at> wrote: > Hi Ken, > The firewall has already been disabled and with the Nemesis channel it worked out of problems.... > Do you maybe know another reason for this issue ? > > Greetings, > Philip > ________________________________________ > Von: discuss-bounces(a)mpich.org [discuss-bounces(a)mpich.org]" im Auftrag von "Kenneth Raffenetti [raffenet(a)mcs.anl.gov] > Gesendet: Dienstag, 8. Juli 2014 14:46 > An: discuss(a)mpich.org > Betreff: Re: [mpich-discuss] Problem with "sock"-channel > > It looks like there may be a firewall blocking connections between > machines. Try disabling it on your hosts and running again. > > Ken > > On 07/08/2014 04:31 AM, Klaus Philip wrote: >> Sorry the pictures where not inserted correctly…. >> >> 1)Error at my self written MPI program: >> >> 2)Error at the Intel MPI Benchmark: >> >> Greetings, >> >> Philip >> >> *Von:*[email protected] [mailto:[email protected]] *Im >> Auftrag von *Klaus Philip >> *Gesendet:* Dienstag, 8. Juli 2014 11:26 >> *An:* discuss(a)mpich.org >> *Betreff:* [mpich-discuss] Problem with "sock"-channel >> >> 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: >> >> >> 2) Error at the Intel MPI Benchmark: >> >> >> So I hope someone of you knows what issue is causing these errors. >> Greetings from Austria, >> Philip >> >> >> >> _______________________________________________ >> 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] Problem with "sock"-channel
by Klaus Philip 08 Jul '14

08 Jul '14

Hi Ken, The firewall has already been disabled and with the Nemesis channel it worked out of problems.... Do you maybe know another reason for this issue ? Greetings, Philip ________________________________________ Von: discuss-bounces(a)mpich.org [discuss-bounces(a)mpich.org]" im Auftrag von "Kenneth Raffenetti [raffenet(a)mcs.anl.gov] Gesendet: Dienstag, 8. Juli 2014 14:46 An: discuss(a)mpich.org Betreff: Re: [mpich-discuss] Problem with "sock"-channel It looks like there may be a firewall blocking connections between machines. Try disabling it on your hosts and running again. Ken On 07/08/2014 04:31 AM, Klaus Philip wrote: > Sorry the pictures where not inserted correctly…. > > 1)Error at my self written MPI program: > > 2)Error at the Intel MPI Benchmark: > > Greetings, > > Philip > > *Von:*[email protected] [mailto:[email protected]] *Im > Auftrag von *Klaus Philip > *Gesendet:* Dienstag, 8. Juli 2014 11:26 > *An:* discuss(a)mpich.org > *Betreff:* [mpich-discuss] Problem with "sock"-channel > > 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: > > > 2) Error at the Intel MPI Benchmark: > > > So I hope someone of you knows what issue is causing these errors. > Greetings from Austria, > Philip > > > > _______________________________________________ > 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] Request‏
by "Antonio J. Peña" 08 Jul '14

08 Jul '14

[Please next time use discuss(a)mpich.org] Sorry, we don't provide support for Windows any more. Microsoft and Intel have Windows versions of their MPI implementations based on MPICH. Please refer to our downloads page: http://www.mpich.org/downloads/ Best, Antonio On 07/08/2014 08:13 AM, Mohammad Nami Nazari wrote: > Hi > I have installed the "mpich2-1.0.1-1-win64-x86-64 on windows 7, 64 bit > OS", but I cant work with it correctly. When I run a parallel program, > face to this error in cmd window: > " Error while connecting to host, No connection could be made because > the target machine actively refused it. <10061> > unable to connect to ' user name:8676' > sock error: error = -1 " > Please help me > > Thanks regard -- Antonio J. Peña Postdoctoral Appointee Mathematics and Computer Science Division Argonne National Laboratory 9700 South Cass Avenue, Bldg. 240, Of. 3148 Argonne, IL 60439-4847 apenya(a)mcs.anl.gov www.mcs.anl.gov/~apenya

1 0

Re: [mpich-discuss] Problem with "sock"-channel
by Kenneth Raffenetti 08 Jul '14

08 Jul '14

It looks like there may be a firewall blocking connections between machines. Try disabling it on your hosts and running again. Ken On 07/08/2014 04:31 AM, Klaus Philip wrote: > Sorry the pictures where not inserted correctly…. > > 1)Error at my self written MPI program: > > 2)Error at the Intel MPI Benchmark: > > Greetings, > > Philip > > *Von:*[email protected] [mailto:[email protected]] *Im > Auftrag von *Klaus Philip > *Gesendet:* Dienstag, 8. Juli 2014 11:26 > *An:* discuss(a)mpich.org > *Betreff:* [mpich-discuss] Problem with "sock"-channel > > 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: > > > 2) Error at the Intel MPI Benchmark: > > > So I hope someone of you knows what issue is causing these errors. > Greetings from Austria, > Philip > > > > _______________________________________________ > 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 with "sock"-channel
by Klaus Philip 08 Jul '14

08 Jul '14

Sorry the pictures where not inserted correctly.... 1) Error at my self written MPI program: [cid:[email protected]] 2) Error at the Intel MPI Benchmark: [cid:[email protected]] Greetings, Philip Von: discuss-bounces(a)mpich.org [mailto:[email protected]] Im Auftrag von Klaus Philip Gesendet: Dienstag, 8. Juli 2014 11:26 An: discuss(a)mpich.org Betreff: [mpich-discuss] Problem with "sock"-channel 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: 2) Error at the Intel MPI Benchmark: So I hope someone of you knows what issue is causing these errors. Greetings from Austria, Philip

1 0