July 2013 - discuss - mailman3.cels.anl.gov

Re: [mpich-discuss] error with MPICH2 while make on Ubuntu 12.04LTS
by Rajeev Thakur 25 Jul '13

25 Jul '13

The logs indicate that configure can run ifort but make cannot, which is strange. On Jul 25, 2013, at 12:26 PM, ZhengCharles wrote: > > > From: charles.zheng(a)msn.com > To: mpich-discuss(a)mcs.anl.gov > Subject: error with MPICH2 while make on Ubuntu 12.04LTS > Date: Fri, 26 Jul 2013 01:22:54 +0800 > > > > From: charles.zheng(a)msn.com > To: mpich-discuss(a)mcs.anl.gov > Subject: error with MPICH2 while make on Ubuntu 12.04LTS > Date: Fri, 26 Jul 2013 01:10:19 +0800 > > Hi, > First of all, thanks for your looking though my mail. > > I'm using the ubuntu 12.04 LTS > > have installed this fortran： > ifort (IFORT) 13.1.0 20130121Package ID: l_fcompxe_2013.2.146 > > This package contains: > Intel(R) Fortran Compiler XE 13.1 for Linux* > Intel(R) Debugger 13.0 for Linux* > Intel(R) Math Kernel Library 11.0 Update 2 for Linux* OS > > and I'm trying to install mpich-3.0.4 on my PC computer. > > I was stuck in the make step. > > When I collect config.logs, I found that I don't have mpich-3.0.4/src/pm/hydra/tools/topo/hwloc/hwloc/config.log > > Using mpich2 to realize parallel in abinit with ifort is my purpose. I'm looking for your reply. > > BEST WITHES > > Sincerely, > > Charles > <txts&logs.tar.bz2>_______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss

1 0

error with MPICH2 while make on Ubuntu 12.04LTS
by ZhengCharles 25 Jul '13

25 Jul '13

From: charles.zheng(a)msn.com To: mpich-discuss(a)mcs.anl.gov Subject: error with MPICH2 while make on Ubuntu 12.04LTS Date: Fri, 26 Jul 2013 01:22:54 +0800 From: charles.zheng(a)msn.com To: mpich-discuss(a)mcs.anl.gov Subject: error with MPICH2 while make on Ubuntu 12.04LTS Date: Fri, 26 Jul 2013 01:10:19 +0800 Hi,First of all, thanks for your looking though my mail. I'm using the ubuntu 12.04 LTS have installed this fortran：ifort (IFORT) 13.1.0 20130121Package ID: l_fcompxe_2013.2.146 This package contains: Intel(R) Fortran Compiler XE 13.1 for Linux* Intel(R) Debugger 13.0 for Linux* Intel(R) Math Kernel Library 11.0 Update 2 for Linux* OS and I'm trying to install mpich-3.0.4 on my PC computer. I was stuck in the make step. When I collect config.logs, I found that I don't have mpich-3.0.4/src/pm/hydra/tools/topo/hwloc/hwloc/config.log Using mpich2 to realize parallel in abinit with ifort is my purpose. I'm looking for your reply. BEST WITHES Sincerely, Charles

1 0

Re: [mpich-discuss] how to specify CFLAGS in configuring mpich
by Syed. Jahanzeb Maqbool Hashmi 24 Jul '13

24 Jul '13

Thanks for the help On Thu, Jul 25, 2013 at 10:21 AM, Pavan Balaji <balaji(a)mcs.anl.gov> wrote: > > You don't need any of these flags if you are building mpich on an armv7 > platform. Just do ./configure. > > -- Pavan > > > On 07/24/2013 07:26 PM, Syed. Jahanzeb Maqbool Hashmi wrote: > >> Hello >> >> Sorry if I am being so noob in asking this. >> >> I want to build mpich for armv7 and in doing so, I want to specify arm >> related flags specifically, CFLAGS="-O2 –march=armv7-a –mtune=cortex-a8 >> –mfpu=neon >> –mfloat-abi=hard –fomit-frame-pointer" >> >> But when I configure mpich and specify CFLAGS it fails to start by >> throwing this error. >> >> linaro@weiser1:/mnt/nfs/**install/mpich-3.0.4$ ./configure >> --prefix=/mnt/nfs/install/**mpich-install/4.8/ CFLAGS="-O2 >> -march=armv7-a -mtune=cortex-a9 -mfpu=neon -float-abi=hard >> -fomit-frame-pointer" >> Configuring MPICH version 3.0.4 with >> '--prefix=/mnt/nfs/install/**mpich-install/4.8/' 'CFLAGS=-O2 >> -march=armv7-a -mtune=cortex-a9 -mfpu=neon -float-abi=hard >> -fomit-frame-pointer' 'CXXFLAGS=-O2 -march=armv7-a >> -mtune=cortex-a9 -mfpu=neon -mfloat-abi=hard >> -funsafe-math-optimizations -ffast-math -fomit-frame-pointer' >> Running on system: Linux weiser1 3.6.0-rc5 #1 SMP PREEMPT Wed >> Sep 12 16:25:07 KST 2012 armv7l armv7l armv7l GNU/Linux >> checking for icc... no >> checking for pgcc... no >> checking for xlc... no >> checking for xlC... no >> checking for pathcc... no >> checking for cc... cc >> checking whether the C compiler works... no >> configure: error: in `/mnt/nfs/install/mpich-3.0.4'**: >> configure: error: C compiler cannot create executables >> See `config.log' for more details >> >> >> Any help? >> >> Thanks >> >> >> ______________________________**_________________ >> discuss mailing list discuss(a)mpich.org >> To manage subscription options or unsubscribe: >> https://lists.mpich.org/**mailman/listinfo/discuss<https://lists.mpich.org/mailman/listinfo/discuss> >> >> > -- > Pavan Balaji > http://www.mcs.anl.gov/~balaji >

1 0

Re: [mpich-discuss] how to specify CFLAGS in configuring mpich
by Pavan Balaji 24 Jul '13

24 Jul '13

You don't need any of these flags if you are building mpich on an armv7 platform. Just do ./configure. -- Pavan On 07/24/2013 07:26 PM, Syed. Jahanzeb Maqbool Hashmi wrote: > Hello > > Sorry if I am being so noob in asking this. > > I want to build mpich for armv7 and in doing so, I want to specify arm > related flags specifically, CFLAGS="-O2 ï¿½march=armv7-a ï¿½mtune=cortex-a8 > ï¿½mfpu=neon > ï¿½mfloat-abi=hard ï¿½fomit-frame-pointer" > > But when I configure mpich and specify CFLAGS it fails to start by > throwing this error. > > linaro@weiser1:/mnt/nfs/install/mpich-3.0.4$ ./configure > --prefix=/mnt/nfs/install/mpich-install/4.8/ CFLAGS="-O2 > -march=armv7-a -mtune=cortex-a9 -mfpu=neon -float-abi=hard > -fomit-frame-pointer" > Configuring MPICH version 3.0.4 with > '--prefix=/mnt/nfs/install/mpich-install/4.8/' 'CFLAGS=-O2 > -march=armv7-a -mtune=cortex-a9 -mfpu=neon -float-abi=hard > -fomit-frame-pointer' 'CXXFLAGS=-O2 -march=armv7-a > -mtune=cortex-a9 -mfpu=neon -mfloat-abi=hard > -funsafe-math-optimizations -ffast-math -fomit-frame-pointer' > Running on system: Linux weiser1 3.6.0-rc5 #1 SMP PREEMPT Wed > Sep 12 16:25:07 KST 2012 armv7l armv7l armv7l GNU/Linux > checking for icc... no > checking for pgcc... no > checking for xlc... no > checking for xlC... no > checking for pathcc... no > checking for cc... cc > checking whether the C compiler works... no > configure: error: in `/mnt/nfs/install/mpich-3.0.4': > configure: error: C compiler cannot create executables > See `config.log' for more details > > > Any help? > > Thanks > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

how to specify CFLAGS in configuring mpich
by Syed. Jahanzeb Maqbool Hashmi 24 Jul '13

24 Jul '13

Hello Sorry if I am being so noob in asking this. I want to build mpich for armv7 and in doing so, I want to specify arm related flags specifically, CFLAGS="-O2 –march=armv7-a –mtune=cortex-a8 –mfpu=neon –mfloat-abi=hard –fomit-frame-pointer" But when I configure mpich and specify CFLAGS it fails to start by throwing this error. linaro@weiser1:/mnt/nfs/install/mpich-3.0.4$ ./configure > --prefix=/mnt/nfs/install/mpich-install/4.8/ CFLAGS="-O2 -march=armv7-a > -mtune=cortex-a9 -mfpu=neon -float-abi=hard -fomit-frame-pointer" > Configuring MPICH version 3.0.4 with > '--prefix=/mnt/nfs/install/mpich-install/4.8/' 'CFLAGS=-O2 -march=armv7-a > -mtune=cortex-a9 -mfpu=neon -float-abi=hard -fomit-frame-pointer' > 'CXXFLAGS=-O2 -march=armv7-a -mtune=cortex-a9 -mfpu=neon -mfloat-abi=hard > -funsafe-math-optimizations -ffast-math -fomit-frame-pointer' > Running on system: Linux weiser1 3.6.0-rc5 #1 SMP PREEMPT Wed Sep 12 > 16:25:07 KST 2012 armv7l armv7l armv7l GNU/Linux > checking for icc... no > checking for pgcc... no > checking for xlc... no > checking for xlC... no > checking for pathcc... no > checking for cc... cc > checking whether the C compiler works... no > configure: error: in `/mnt/nfs/install/mpich-3.0.4': > configure: error: C compiler cannot create executables > See `config.log' for more details Any help? Thanks

1 0

Re: [mpich-discuss] MPI_Scatter message truncated
by Pavan Balaji 24 Jul '13

24 Jul '13

Please send us a simple program that reproduces this problem. -- Pavan On 07/24/2013 04:21 PM, Sufeng Niu wrote: > Hi, > > I apologize to post so many questions, but I got a new question here: my > program runs everything well on 1 machine right now. when I run it on > two machines, the program got error: > > Fatal error in PMPI_Scatter: Message truncated, > MPIDI_CH3U_Receive_data_found(131) Message from rank 0 and tag 5 > truncated. 1774000 bytes received but buffer size is 52. > MPIC_Send(59).......: Negative count, value is -32 > > I cannot understand how does this happened. the hello world is tested > for network link. and the program is fine running on one computer. > > Thank you very much for any suggestions > > -- > Best Regards, > Sufeng Niu > ECASP lab, ECE department, Illinois Institute of Technology > Tel: 312-731-7219 > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

MPI_Scatter message truncated
by Sufeng Niu 24 Jul '13

24 Jul '13

Hi, I apologize to post so many questions, but I got a new question here: my program runs everything well on 1 machine right now. when I run it on two machines, the program got error: Fatal error in PMPI_Scatter: Message truncated, MPIDI_CH3U_Receive_data_found(131) Message from rank 0 and tag 5 truncated. 1774000 bytes received but buffer size is 52. MPIC_Send(59).......: Negative count, value is -32 I cannot understand how does this happened. the hello world is tested for network link. and the program is fine running on one computer. Thank you very much for any suggestions -- Best Regards, Sufeng Niu ECASP lab, ECE department, Illinois Institute of Technology Tel: 312-731-7219

1 0

Default behavior of mpif90/ifort on MacOSX (MH_BUNDLE, MH_DYLIB)
by Bodony, Daniel J 24 Jul '13

24 Jul '13

Hello, After upgrading to mpich-3.0.4 today I was presented with the following error on my MacOSX 10.8.4 machine when using mpif90 (installation/compiler info at e-mail bottom). For the simple code: > [02:23 PM Wed Jul 24] > cat main.f90 > Program MAIN > > Print *, "F: Hello, World!" > > End Program MAIN The compilation gives > [02:26 PM Wed Jul 24] > mpif90 main.f90 > ld: can't link with bundle (MH_BUNDLE) only dylibs (MH_DYLIB) for architecture x86_64 and no executable is built. Using mpif90 -link-info and chasing down the libraries I found that the bundle libmpichf90.so was being linked instead of libmpichf90.dylib (which is not created) or the static library libmpichf90.a. (That libmpichf90.so was a bundle was confirmed using otool -hV /usr/local/lib/libmpichf90.so.) After hiding the library libmpichf90.so so the static version libmpichf90.a is used, compile proceeds normally. This behavior was not seen earlier when using mpich2-1.4.1 through mpich-3.0.1 (I think). COMPILER INFO: > [02:36 PM Wed Jul 24] > ifort -V > Intel(R) Fortran Intel(R) 64 Compiler XE for applications running on Intel(R) 64, Version 13.0.3.198 Build 20130606 > > [02:36 PM Wed Jul 24] > icc -V > Intel(R) C Intel(R) 64 Compiler XE for applications running on Intel(R) 64, Version 13.0.3.198 Build 20130606 > > [02:36 PM Wed Jul 24] > icpc -V > Intel(R) C++ Intel(R) 64 Compiler XE for applications running on Intel(R) 64, Version 13.0.3.198 Build 20130606 MPICH INFO: (build used ifort/icc/icpc as above) > [02:38 PM Wed Jul 24] > head config.log > This file contains any messages produced by compilers while > running configure, to aid debugging if configure makes a mistake. > > It was created by MPICH configure 3.0.4, which was > generated by GNU Autoconf 2.69. Invocation command line was > > $ ./configure --enable-f77 --enable-fc --enable-cxx --enable-shared --enable-romio --enable-mpe Cheers, Dan -- Daniel J. Bodony Associate Professor of Aerospace Engineering Affiliate Professor of Mechanical Science & Engineering University of Illinois at Urbana-Champaign

1 0

Re: [mpich-discuss] MPI_Win_free failed
by Sufeng Niu 24 Jul '13

24 Jul '13

Hi Dr. Pavan, Thank you so much for your reply. Sorry to post the whole code. I found out the where is the problem. due to my different communicator, 1 process doesn't create window. so I failed to free it. Once again. Thanks a lot! Sufeng On Tue, Jul 23, 2013 at 6:27 PM, Pavan Balaji <balaji(a)mcs.anl.gov> wrote: > > Please send a *simple* example which demonstrates the problem. > > I'm almost certain it's a problem with your application and not with MPI. > Hence you going through the exercise of creating a simple example will > allow you to figure out what's wrong. > > -- Pavan > > > On 07/23/2013 06:10 PM, Sufeng Niu wrote: > >> sure, no problem. but I guess svn maybe easier. since there is some >> images file to run it. Can you use svn to clone it? >> >> it is: https://subversion.xray.aps.**anl.gov/detpool/SUFENGNIU/MPI_** >> Compression<https://subversion.xray.aps.anl.gov/detpool/SUFENGNIU/MPI_Compression> >> >> just make, you need two terminals to run it. one terminal type: >> ./run.perl main >> the second terminal type: ./udp_client 55Fe_run5.tif >> >> which is that the first terminal send the data, the second terminal >> receive it and process it. >> >> Thank you so much. Due to my limited knowledge I didn't find out where >> is the problem. >> >> once again, thanks a lot! >> >> Sufeng >> >> >> On Tue, Jul 23, 2013 at 6:00 PM, Pavan Balaji <balaji(a)mcs.anl.gov >> <mailto:[email protected]>> wrote: >> >> >> Can you send us a simple program that gives this error? >> >> -- Pavan >> >> >> On 07/23/2013 05:55 PM, Sufeng Niu wrote: >> >> Hello, >> >> I used MPI_Win_free, and it returns error code: 45, Invalid >> MPI_Win. I >> checked all my code, but I still cannot find out where is the >> problem. I >> was wondering any one can give a suggestion? >> >> I defined win as >> MPI_Win win; >> >> in RMA, I create a sub communicator: MPI_FOR_WORLD >> MPI_Win_create(image_buff, >> 2*image_info->image_size*__**sizeof(uint16), >> >> sizeof(uint16), MPI_INFO_NULL, MPI_FOR_WORLD, win); >> >> In other MPI_FOR_WORLD process: >> MPI_Win_create(MPI_BOTTOM, 0, sizeof(uint16), MPI_INFO_NULL, >> MPI_FOR_WORLD, win); >> >> I use MPI_Get, just as normal thing. no mem violation. >> >> free it as >> MPI_Win_free(&win); >> >> I still cannot figure out why it keeps saying MPI_Win_free >> failed. Thank >> you very much for any suggestions. >> >> -- >> Best Regards, >> Sufeng Niu >> ECASP lab, ECE department, Illinois Institute of Technology >> Tel: 312-731-7219 <tel:312-731-7219> <tel:312-731-7219 >> <tel:312-731-7219>> >> >> >> ______________________________**___________________ >> discuss mailing list discuss(a)mpich.org <mailto:[email protected]> >> >> To manage subscription options or unsubscribe: >> https://lists.mpich.org/__**mailman/listinfo/discuss<https://lists.mpich.org/__mailman/listinfo/discuss> >> >> <https://lists.mpich.org/**mailman/listinfo/discuss<https://lists.mpich.org/mailman/listinfo/discuss> >> > >> >> >> -- >> Pavan Balaji >> http://www.mcs.anl.gov/~balaji >> >> >> >> >> -- >> Best Regards, >> Sufeng Niu >> ECASP lab, ECE department, Illinois Institute of Technology >> Tel: 312-731-7219 >> > > -- > Pavan Balaji > http://www.mcs.anl.gov/~balaji > -- Best Regards, Sufeng Niu ECASP lab, ECE department, Illinois Institute of Technology Tel: 312-731-7219

1 0

Re: [mpich-discuss] hydra crashes with high number of processes
by Pavan Balaji 24 Jul '13

24 Jul '13

On 07/24/2013 10:30 AM, Antonio J. Peï¿½a wrote: > Hi Thomas, > > Thanks for reporting. Could you please give a try with the current > development version and let us know if you still experience the segfault? > > git clone git://git.mpich.org/mpich.git > > OR alternatively > > http://git.mpich.org/mpich.git Perhaps better to use the nightly tarballs: http://www.mpich.org/static/tarballs/nightly/master/ -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0