- discuss - mailman3.cels.anl.gov

Re: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count
by Zhou, Hui 26 Mar '24

26 Mar '24

Hi Iker, Could you try FI_PROVIDER=verbs? -- Hui ________________________________ From: Iker Martín Álvarez <martini(a)uji.es> Sent: Tuesday, March 26, 2024 2:35 AM To: Zhou, Hui <zhouh(a)anl.gov> Cc: discuss(a)mpich.org <discuss(a)mpich.org> Subject: Re: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count Hello Zhou, I just tried using both providers and in both cases the execution hangs at MPI_Init for those processes that are on a different node than the main one. This happens even for an MPI "Hello World" code for both providers. So ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd Hello Zhou, I just tried using both providers and in both cases the execution hangs at MPI_Init for those processes that are on a different node than the main one. This happens even for an MPI "Hello World" code for both providers. So I can't even check if it has the same problem. Thank you very much for your time. Best regards, Iker Missatge de Zhou, Hui <zhouh(a)anl.gov<mailto:[email protected]>> del dia dl., 25 de març 2024 a les 23:05: Hi Iker, Could you try setting FI_PROVIDER=sockets or FI_PROVIDER=tcp to see if the issue persists? -- Hui ________________________________ From: Iker Martín Álvarez <martini(a)uji.es<mailto:[email protected]>> Sent: Monday, March 25, 2024 12:38 PM To: Zhou, Hui <zhouh(a)anl.gov<mailto:[email protected]>> Cc: discuss(a)mpich.org<mailto:[email protected]> <discuss(a)mpich.org<mailto:[email protected]>> Subject: Re: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count Hello Zhou, Thanks for the quick reply. In the attached file you can see the result of running the code with the environment variable you gave. Kind regards, Iker Missatge de Zhou, Hui <zhouh@ anl. gov> del dia dl. , 25 de març 2024 a les ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd Hello Zhou, Thanks for the quick reply. In the attached file you can see the result of running the code with the environment variable you gave. Kind regards, Iker Missatge de Zhou, Hui <zhouh(a)anl.gov<mailto:[email protected]>> del dia dl., 25 de març 2024 a les 17:03: Hi Iker, Could you try reproduce the issue by setting MPIR_CVAR_DEBUG_SUMMARY=1, and report the console output? The issue may be in specific provider. The log should show that. -- Hui ________________________________ From: Iker Martín Álvarez via discuss <discuss(a)mpich.org<mailto:[email protected]>> Sent: Monday, March 25, 2024 6:23 AM To: discuss(a)mpich.org<mailto:[email protected]> <discuss(a)mpich.org<mailto:[email protected]>> Cc: Iker Martín Álvarez <martini(a)uji.es<mailto:[email protected]>> Subject: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count Hello, I recently encountered an unexpected behaviour of the MPI_Probe + MPI_Get_count functions under specific conditions. I was hoping that this forum could advise me on a solution. Specifically, the application performs an MPI_Send communication ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd Hello, I recently encountered an unexpected behaviour of the MPI_Probe + MPI_Get_count functions under specific conditions. I was hoping that this forum could advise me on a solution. Specifically, the application performs an MPI_Send communication from the root process to process B. Process B doesn't know the size of the message, so I use MPI_Probe + MPI_Get_count to discover it. However, as an example, if the size of the message is 1000 bytes, process B expects with MPI_Get_count function a total of 20 bytes. The problem only occurs with a specific installation of MPICH and when the following conditions are met in my code: - The problem only occurs in internode communications. - The problem only appears if derived types are used in the communication. Specifically a derived type to communicate a vector of integers and a vector of reals, both with the same number of elements. - None of the MPI functions give an error code. They all return MPI_Sucess. - If instead of allocating the amount of bytes returned by MPI_Get_count(=20), I allocate the expected value (1000), the message is received correctly. - The size returned by MPI_Get_count seems to be variable depending on the total number of addresses with which the derived type is created. I have attached the file to reproduce the problem. It can also be accessed via the GitLab link below: https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of… <https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of…> It is designed to be run with 3 processes, two of them hosted on one node and the third on a different one. As previously mentioned, this problem occurs when using MPICH with ch4:ofi without using the embedded option. Specifically, I have tested the following installations in which the error appears: - MPICH 4.2.0 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' '--disable-psm3' - MPICH 3.4.1 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' However, it does work as expected for the following MPICH installations: - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=embedded' - MPICH 4.0.3 with config options: '--with-device=ch4:ucx' '--with-ucx=/soft/gnu/ucx-1.11' - MPICH 3.4.1 with config options: '--with-device=ch4:ucx' '--with-ucx=/soft/gnu/ucx-1.11' Although for these installations the code does work, we would like to use a different libfabric installation than the embedded one because we get better networking performance. In the case of UCX, it is because the application in question uses the MPI_Comm_spawn call and MPICH does not currently support it with UCX. Thank you for your help. Best regards, Iker

1 0

Help with MPICH 4.2.0 and win_allocate_shared (or maybe infiniband?)
by Thompson, Matt (GSFC-610.1)[SCIENCE SYSTEMS AND APPLICATIONS INC] 26 Mar '24

26 Mar '24

All, I've been trying to get a code of mine working with MPICH 4.2.0. I can build MPICH just fine and then build our base libraries and then model and all compiles fine. Hello world runs fine on multiple nodes as well. But when I finally try and run our complex model: Assertion failed in file ../src/mpid/ch4/src/ch4_impl.h at line 838: map_entry != NULL /discover/swdev/gmao_SIteam/MPI/mpich/4.2.0-SLES15/ifort-2021.11.0/lib/libmpi.so.12(+0x37d211) [0x14bf4f62c211] /discover/swdev/gmao_SIteam/MPI/mpich/4.2.0-SLES15/ifort-2021.11.0/lib/libmpi.so.12(PMPI_Win_allocate_shared+0x3ba) [0x14bf4f3e452a] /discover/swdev/gmao_SIteam/Baselibs/ESMA-Baselibs-7.20.0/x86_64-pc-linux-gnu/ifort_2021.11.0-mpich_4.2.0-SLES15/Linux/lib/libesmf.so(_ZN5ESMCI3VMK14ssishmAllocateERSt6vectorImSaImEEPNS0_9memhandleEb+0x18b) [0x14bf6e91481b] /discover/swdev/gmao_SIteam/Baselibs/ESMA-Baselibs-7.20.0/x86_64-pc-linux-gnu/ifort_2021.11.0-mpich_4.2.0-SLES15/Linux/lib/libesmf.so(_ZN5ESMCI5Array6createEPNS_9ArraySpecEPNS_8DistGridEPNS_10InterArrayIiEES7_S7_S7_S7_S7_S7_P14ESMC_IndexFlagP13ESMC_Pin_FlagS7_S7_S7_PiPNS_2VME+0x2707) [0x14bf6e44a267] What I'm mainly wondering is if anyone has any experience with an error like this? My guess (at the moment) is that I built things wrong for an Infiniband cluster maybe? I'm using Intel Fortran Classic 2021.11.0 with GCC 11.4.0 as a backing C compiler and I built as: mkdir build-ifort-2021.11.0 && cd build-ifort-2021.11.0 ../configure CC=icx CXX=icpx FC=ifort \ --with-ucx=embedded --with-hwloc=embedded --with-libfabric=embedded --with-yaksa=embedded \ --prefix=/discover/swdev/gmao_SIteam/MPI/mpich/4.2.0-SLES15/ifort-2021.11.0 |& tee configure.ifort-2021.11.0.log All those "embedded" flags are mainly because with Open MPI on this system, I have to do something similar with its configure step: --with-hwloc=internal --with-libevent=internal --with-pmix=internal so I figured I should do the same with MPICH> Now, at the end of the configure step I did see: ***************************************************** *** *** device : ch4:ofi (embedded libfabric) *** shm feature : auto *** gpu support : disabled *** MPICH is configured with device ch4:ofi, which should work for TCP networks and any high-bandwidth interconnect supported by libfabric. MPICH can also be configured with "--with-device=ch4:ucx", which should work for TCP networks and any high-bandwidth interconnect supported by the UCX library. In addition, the legacy device ch3 (--with-device=ch3) is also available. ***************************************************** And I did try the `--with-device=ch4:ucx` but that didn't seem to help. And the system I am on is an Infiniband network, so I imagine ofi should work. Note that this code works fine with Intel MPI and Open MPI (which are our "main" MPI stacks), but some of our external users are asking about MPICH support. Matt -- Matt Thompson, SSAI, Ld Scientific Prog/Analyst/Super NASA GSFC, Global Modeling and Assimilation Office Code 610.1, 8800 Greenbelt Rd, Greenbelt, MD 20771 Phone: 301-614-6712 Fax: 301-614-6246 https://urldefense.us/v3/__http://science.gsfc.nasa.gov/sed/bio/matthew.tho…

1 0

Re: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count
by Iker Martín Álvarez 26 Mar '24

26 Mar '24

Hello Zhou, I just tried using both providers and in both cases the execution hangs at MPI_Init for those processes that are on a different node than the main one. This happens even for an MPI "Hello World" code for both providers. So I can't even check if it has the same problem. Thank you very much for your time. Best regards, Iker Missatge de Zhou, Hui <zhouh(a)anl.gov> del dia dl., 25 de març 2024 a les 23:05: > Hi Iker, > > Could you try setting FI_PROVIDER=sockets or FI_PROVIDER=tcp to see if > the issue persists? > > -- > Hui > ------------------------------ > *From:* Iker Martín Álvarez <martini(a)uji.es> > *Sent:* Monday, March 25, 2024 12:38 PM > *To:* Zhou, Hui <zhouh(a)anl.gov> > *Cc:* discuss(a)mpich.org <discuss(a)mpich.org> > *Subject:* Re: [mpich-discuss] Unexpected behaviour of MPI_Probe + > MPI_Get_count > > Hello Zhou, Thanks for the quick reply. In the attached file you can see > the result of running the code with the environment variable you gave. Kind > regards, Iker Missatge de Zhou, Hui <zhouh@ anl. gov> del dia dl. , 25 de > març 2024 a les > ZjQcmQRYFpfptBannerStart > This Message Is From an External Sender > This message came from outside your organization. > > ZjQcmQRYFpfptBannerEnd > Hello Zhou, > > Thanks for the quick reply. > In the attached file you can see the result of running the code with the > environment variable you gave. > > Kind regards, > Iker > > Missatge de Zhou, Hui <zhouh(a)anl.gov> del dia dl., 25 de març 2024 a les > 17:03: > > Hi Iker, > > Could you try reproduce the issue by setting MPIR_CVAR_DEBUG_SUMMARY=1, > and report the console output? The issue may be in specific provider. The > log should show that. > > -- > Hui > ------------------------------ > *From:* Iker Martín Álvarez via discuss <discuss(a)mpich.org> > *Sent:* Monday, March 25, 2024 6:23 AM > *To:* discuss(a)mpich.org <discuss(a)mpich.org> > *Cc:* Iker Martín Álvarez <martini(a)uji.es> > *Subject:* [mpich-discuss] Unexpected behaviour of MPI_Probe + > MPI_Get_count > > Hello, I recently encountered an unexpected behaviour of the MPI_Probe + > MPI_Get_count functions under specific conditions. I was hoping that this > forum could advise me on a solution. Specifically, the application > performs an MPI_Send communication > ZjQcmQRYFpfptBannerStart > This Message Is From an External Sender > This message came from outside your organization. > > ZjQcmQRYFpfptBannerEnd > Hello, > > I recently encountered an unexpected behaviour of the MPI_Probe + > MPI_Get_count functions under specific conditions. I was hoping that this > forum could advise me on a solution. > > Specifically, the application performs an MPI_Send communication from the > root process to process B. Process B doesn't know the size of the message, > so I use MPI_Probe + MPI_Get_count to discover it. However, as an example, > if the size of the message is 1000 bytes, process B expects with > MPI_Get_count function a total of 20 bytes. > > The problem only occurs with a specific installation of MPICH and when the > following conditions are met in my code: > - The problem only occurs in internode communications. > - The problem only appears if derived types are used in the communication. > Specifically a derived type to communicate a vector of integers and a > vector of reals, both with the same number of elements. > - None of the MPI functions give an error code. They all return MPI_Sucess. > - If instead of allocating the amount of bytes returned by > MPI_Get_count(=20), I allocate the expected value (1000), the message is > received correctly. > - The size returned by MPI_Get_count seems to be variable depending on the > total number of addresses with which the derived type is created. > > I have attached the file to reproduce the problem. It can also be accessed > via the GitLab link below: > https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of… > <https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of…> > It is designed to be run with 3 processes, two of them hosted on one node > and the third on a different one. > > As previously mentioned, this problem occurs when using MPICH with ch4:ofi > without using the embedded option. Specifically, I have tested the > following installations in which the error appears: > - MPICH 4.2.0 with config options: '--with-device=ch4:ofi' > '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' > - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' > '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' > - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' > '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' '--disable-psm3' > - MPICH 3.4.1 with config options: '--with-device=ch4:ofi' > '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' > > However, it does work as expected for the following MPICH installations: > - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' > '--with-libfabric=embedded' > - MPICH 4.0.3 with config options: '--with-device=ch4:ucx' > '--with-ucx=/soft/gnu/ucx-1.11' > - MPICH 3.4.1 with config options: '--with-device=ch4:ucx' > '--with-ucx=/soft/gnu/ucx-1.11' > > Although for these installations the code does work, we would like to use > a different libfabric installation than the embedded one because we get > better networking performance. In the case of UCX, it is because the > application in question uses the MPI_Comm_spawn call and MPICH does not > currently support it with UCX. > > Thank you for your help. > Best regards, > Iker > >

1 0

Re: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count
by Zhou, Hui 25 Mar '24

25 Mar '24

Hi Iker, Could you try setting FI_PROVIDER=sockets or FI_PROVIDER=tcp to see if the issue persists? -- Hui ________________________________ From: Iker Martín Álvarez <martini(a)uji.es> Sent: Monday, March 25, 2024 12:38 PM To: Zhou, Hui <zhouh(a)anl.gov> Cc: discuss(a)mpich.org <discuss(a)mpich.org> Subject: Re: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count Hello Zhou, Thanks for the quick reply. In the attached file you can see the result of running the code with the environment variable you gave. Kind regards, Iker Missatge de Zhou, Hui <zhouh@ anl. gov> del dia dl. , 25 de març 2024 a les ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd Hello Zhou, Thanks for the quick reply. In the attached file you can see the result of running the code with the environment variable you gave. Kind regards, Iker Missatge de Zhou, Hui <zhouh(a)anl.gov<mailto:[email protected]>> del dia dl., 25 de març 2024 a les 17:03: Hi Iker, Could you try reproduce the issue by setting MPIR_CVAR_DEBUG_SUMMARY=1, and report the console output? The issue may be in specific provider. The log should show that. -- Hui ________________________________ From: Iker Martín Álvarez via discuss <discuss(a)mpich.org<mailto:[email protected]>> Sent: Monday, March 25, 2024 6:23 AM To: discuss(a)mpich.org<mailto:[email protected]> <discuss(a)mpich.org<mailto:[email protected]>> Cc: Iker Martín Álvarez <martini(a)uji.es<mailto:[email protected]>> Subject: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count Hello, I recently encountered an unexpected behaviour of the MPI_Probe + MPI_Get_count functions under specific conditions. I was hoping that this forum could advise me on a solution. Specifically, the application performs an MPI_Send communication ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd Hello, I recently encountered an unexpected behaviour of the MPI_Probe + MPI_Get_count functions under specific conditions. I was hoping that this forum could advise me on a solution. Specifically, the application performs an MPI_Send communication from the root process to process B. Process B doesn't know the size of the message, so I use MPI_Probe + MPI_Get_count to discover it. However, as an example, if the size of the message is 1000 bytes, process B expects with MPI_Get_count function a total of 20 bytes. The problem only occurs with a specific installation of MPICH and when the following conditions are met in my code: - The problem only occurs in internode communications. - The problem only appears if derived types are used in the communication. Specifically a derived type to communicate a vector of integers and a vector of reals, both with the same number of elements. - None of the MPI functions give an error code. They all return MPI_Sucess. - If instead of allocating the amount of bytes returned by MPI_Get_count(=20), I allocate the expected value (1000), the message is received correctly. - The size returned by MPI_Get_count seems to be variable depending on the total number of addresses with which the derived type is created. I have attached the file to reproduce the problem. It can also be accessed via the GitLab link below: https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of… <https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of…> It is designed to be run with 3 processes, two of them hosted on one node and the third on a different one. As previously mentioned, this problem occurs when using MPICH with ch4:ofi without using the embedded option. Specifically, I have tested the following installations in which the error appears: - MPICH 4.2.0 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' '--disable-psm3' - MPICH 3.4.1 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' However, it does work as expected for the following MPICH installations: - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=embedded' - MPICH 4.0.3 with config options: '--with-device=ch4:ucx' '--with-ucx=/soft/gnu/ucx-1.11' - MPICH 3.4.1 with config options: '--with-device=ch4:ucx' '--with-ucx=/soft/gnu/ucx-1.11' Although for these installations the code does work, we would like to use a different libfabric installation than the embedded one because we get better networking performance. In the case of UCX, it is because the application in question uses the MPI_Comm_spawn call and MPICH does not currently support it with UCX. Thank you for your help. Best regards, Iker

1 0

Re: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count
by Iker Martín Álvarez 25 Mar '24

25 Mar '24

Hello Zhou, Thanks for the quick reply. In the attached file you can see the result of running the code with the environment variable you gave. Kind regards, Iker Missatge de Zhou, Hui <zhouh(a)anl.gov> del dia dl., 25 de març 2024 a les 17:03: > Hi Iker, > > Could you try reproduce the issue by setting MPIR_CVAR_DEBUG_SUMMARY=1, > and report the console output? The issue may be in specific provider. The > log should show that. > > -- > Hui > ------------------------------ > *From:* Iker Martín Álvarez via discuss <discuss(a)mpich.org> > *Sent:* Monday, March 25, 2024 6:23 AM > *To:* discuss(a)mpich.org <discuss(a)mpich.org> > *Cc:* Iker Martín Álvarez <martini(a)uji.es> > *Subject:* [mpich-discuss] Unexpected behaviour of MPI_Probe + > MPI_Get_count > > Hello, I recently encountered an unexpected behaviour of the MPI_Probe + > MPI_Get_count functions under specific conditions. I was hoping that this > forum could advise me on a solution. Specifically, the application > performs an MPI_Send communication > ZjQcmQRYFpfptBannerStart > This Message Is From an External Sender > This message came from outside your organization. > > ZjQcmQRYFpfptBannerEnd > Hello, > > I recently encountered an unexpected behaviour of the MPI_Probe + > MPI_Get_count functions under specific conditions. I was hoping that this > forum could advise me on a solution. > > Specifically, the application performs an MPI_Send communication from the > root process to process B. Process B doesn't know the size of the message, > so I use MPI_Probe + MPI_Get_count to discover it. However, as an example, > if the size of the message is 1000 bytes, process B expects with > MPI_Get_count function a total of 20 bytes. > > The problem only occurs with a specific installation of MPICH and when the > following conditions are met in my code: > - The problem only occurs in internode communications. > - The problem only appears if derived types are used in the communication. > Specifically a derived type to communicate a vector of integers and a > vector of reals, both with the same number of elements. > - None of the MPI functions give an error code. They all return MPI_Sucess. > - If instead of allocating the amount of bytes returned by > MPI_Get_count(=20), I allocate the expected value (1000), the message is > received correctly. > - The size returned by MPI_Get_count seems to be variable depending on the > total number of addresses with which the derived type is created. > > I have attached the file to reproduce the problem. It can also be accessed > via the GitLab link below: > https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of… > <https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of…> > It is designed to be run with 3 processes, two of them hosted on one node > and the third on a different one. > > As previously mentioned, this problem occurs when using MPICH with ch4:ofi > without using the embedded option. Specifically, I have tested the > following installations in which the error appears: > - MPICH 4.2.0 with config options: '--with-device=ch4:ofi' > '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' > - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' > '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' > - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' > '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' '--disable-psm3' > - MPICH 3.4.1 with config options: '--with-device=ch4:ofi' > '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' > > However, it does work as expected for the following MPICH installations: > - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' > '--with-libfabric=embedded' > - MPICH 4.0.3 with config options: '--with-device=ch4:ucx' > '--with-ucx=/soft/gnu/ucx-1.11' > - MPICH 3.4.1 with config options: '--with-device=ch4:ucx' > '--with-ucx=/soft/gnu/ucx-1.11' > > Although for these installations the code does work, we would like to use > a different libfabric installation than the embedded one because we get > better networking performance. In the case of UCX, it is because the > application in question uses the MPI_Comm_spawn call and MPICH does not > currently support it with UCX. > > Thank you for your help. > Best regards, > Iker >

1 0

Re: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count
by Zhou, Hui 25 Mar '24

25 Mar '24

Hi Iker, Could you try reproduce the issue by setting MPIR_CVAR_DEBUG_SUMMARY=1, and report the console output? The issue may be in specific provider. The log should show that. -- Hui ________________________________ From: Iker Martín Álvarez via discuss <discuss(a)mpich.org> Sent: Monday, March 25, 2024 6:23 AM To: discuss(a)mpich.org <discuss(a)mpich.org> Cc: Iker Martín Álvarez <martini(a)uji.es> Subject: [mpich-discuss] Unexpected behaviour of MPI_Probe + MPI_Get_count Hello, I recently encountered an unexpected behaviour of the MPI_Probe + MPI_Get_count functions under specific conditions. I was hoping that this forum could advise me on a solution. Specifically, the application performs an MPI_Send communication ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd Hello, I recently encountered an unexpected behaviour of the MPI_Probe + MPI_Get_count functions under specific conditions. I was hoping that this forum could advise me on a solution. Specifically, the application performs an MPI_Send communication from the root process to process B. Process B doesn't know the size of the message, so I use MPI_Probe + MPI_Get_count to discover it. However, as an example, if the size of the message is 1000 bytes, process B expects with MPI_Get_count function a total of 20 bytes. The problem only occurs with a specific installation of MPICH and when the following conditions are met in my code: - The problem only occurs in internode communications. - The problem only appears if derived types are used in the communication. Specifically a derived type to communicate a vector of integers and a vector of reals, both with the same number of elements. - None of the MPI functions give an error code. They all return MPI_Sucess. - If instead of allocating the amount of bytes returned by MPI_Get_count(=20), I allocate the expected value (1000), the message is received correctly. - The size returned by MPI_Get_count seems to be variable depending on the total number of addresses with which the derived type is created. I have attached the file to reproduce the problem. It can also be accessed via the GitLab link below: https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of… <https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of…> It is designed to be run with 3 processes, two of them hosted on one node and the third on a different one. As previously mentioned, this problem occurs when using MPICH with ch4:ofi without using the embedded option. Specifically, I have tested the following installations in which the error appears: - MPICH 4.2.0 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' '--disable-psm3' - MPICH 3.4.1 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' However, it does work as expected for the following MPICH installations: - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=embedded' - MPICH 4.0.3 with config options: '--with-device=ch4:ucx' '--with-ucx=/soft/gnu/ucx-1.11' - MPICH 3.4.1 with config options: '--with-device=ch4:ucx' '--with-ucx=/soft/gnu/ucx-1.11' Although for these installations the code does work, we would like to use a different libfabric installation than the embedded one because we get better networking performance. In the case of UCX, it is because the application in question uses the MPI_Comm_spawn call and MPICH does not currently support it with UCX. Thank you for your help. Best regards, Iker

1 0

Unexpected behaviour of MPI_Probe + MPI_Get_count
by Iker Martín Álvarez 25 Mar '24

25 Mar '24

Hello, I recently encountered an unexpected behaviour of the MPI_Probe + MPI_Get_count functions under specific conditions. I was hoping that this forum could advise me on a solution. Specifically, the application performs an MPI_Send communication from the root process to process B. Process B doesn't know the size of the message, so I use MPI_Probe + MPI_Get_count to discover it. However, as an example, if the size of the message is 1000 bytes, process B expects with MPI_Get_count function a total of 20 bytes. The problem only occurs with a specific installation of MPICH and when the following conditions are met in my code: - The problem only occurs in internode communications. - The problem only appears if derived types are used in the communication. Specifically a derived type to communicate a vector of integers and a vector of reals, both with the same number of elements. - None of the MPI functions give an error code. They all return MPI_Sucess. - If instead of allocating the amount of bytes returned by MPI_Get_count(=20), I allocate the expected value (1000), the message is received correctly. - The size returned by MPI_Get_count seems to be variable depending on the total number of addresses with which the derived type is created. I have attached the file to reproduce the problem. It can also be accessed via the GitLab link below: https://urldefense.us/v3/__https://lorca.act.uji.es/gitlab/martini/mpich_of… It is designed to be run with 3 processes, two of them hosted on one node and the third on a different one. As previously mentioned, this problem occurs when using MPICH with ch4:ofi without using the embedded option. Specifically, I have tested the following installations in which the error appears: - MPICH 4.2.0 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' '--disable-psm3' - MPICH 3.4.1 with config options: '--with-device=ch4:ofi' '--with-libfabric=/home/martini/Instalaciones/libfabric-1.16.1' However, it does work as expected for the following MPICH installations: - MPICH 4.0.3 with config options: '--with-device=ch4:ofi' '--with-libfabric=embedded' - MPICH 4.0.3 with config options: '--with-device=ch4:ucx' '--with-ucx=/soft/gnu/ucx-1.11' - MPICH 3.4.1 with config options: '--with-device=ch4:ucx' '--with-ucx=/soft/gnu/ucx-1.11' Although for these installations the code does work, we would like to use a different libfabric installation than the embedded one because we get better networking performance. In the case of UCX, it is because the application in question uses the MPI_Comm_spawn call and MPICH does not currently support it with UCX. Thank you for your help. Best regards, Iker

1 0

Re: [mpich-discuss] Fwd: Trouble with installation of MPICH-3.0.4
by Zhou, Hui 15 Mar '24

15 Mar '24

Hi Dilip, MPICH 3.0.4 was released in 2013. The current MPICH release is 4.2.0 (https://urldefense.us/v3/__https://www.mpich.org/downloads/__;!!G_uCfscf7eW… ). Is there any reason you need to install mpich 3.0.4? Otherwise, we suggest you try mpich-4.2.0. The issue is due to modern Fortran compilers being more strict in type checking. There is a compilation flag (FFLAGS=-fallow-argument-mismatch) that you may try to bypass the issue. But try the latest MPICH release first. Regards, Hui ________________________________ From: Dilip V via discuss <discuss(a)mpich.org> Sent: Friday, March 15, 2024 11:02 AM To: discuss(a)mpich.org <discuss(a)mpich.org> Cc: Dilip V <v.dilip(a)students.iiserpune.ac.in> Subject: [mpich-discuss] Fwd: Trouble with installation of MPICH-3.0.4 Hello, I'm trying to install MPICH(mpich-3. 0. 4. tar. gz) on my Dell Dekstop running on Ubuntu 20. 04 LTS. I'm encountering error message during ./configure that reads --------------------------------------------------------------------------------------------"checking ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. ZjQcmQRYFpfptBannerEnd Hello, I'm trying to install MPICH(mpich-3.0.4.tar.gz) on my Dell Dekstop running on Ubuntu 20.04 LTS. I'm encountering error message during ./configure that reads -------------------------------------------------------------------------------------------- "checking whether gfortran allows mismatched arguments... no configure: error: The Fortran compiler gfortran will not compile files that call the same routine with arguments of different types." ---------------------------------------------------------------------------------------------- I'm using gfortran for my FORTRAN 90 as well as FORTRAN 77 compiler. How do I resolve this? My c.txt is attached below for reference.

1 0

Fwd: Trouble with installation of MPICH-3.0.4
by Dilip V 15 Mar '24

15 Mar '24

Hello, I'm trying to install MPICH(mpich-3.0.4.tar.gz) on my Dell Dekstop running on Ubuntu 20.04 LTS. I'm encountering error message during ./configure that reads -------------------------------------------------------------------------------------------- "checking whether gfortran allows mismatched arguments... no configure: error: The Fortran compiler gfortran will not compile files that call the same routine with arguments of different types." ---------------------------------------------------------------------------------------------- I'm using gfortran for my FORTRAN 90 as well as FORTRAN 77 compiler. How do I resolve this? My c.txt is attached below for reference.

1 0

Re: [mpich-discuss] Install errors on MAC OS Sonoma: "error: expected declaration specifiers or ‘...’ before string constant"
by Douglas G Dommermuth 02 Mar '24

02 Mar '24

I tried a different channel as you had suggested: ./configure --prefix=/Users/douglasdommermuth/mpich2/mpich2-install --with-device=ch3:nemesis |& tee c.txt As a result, I was able to install mpich. Thank you, Doug. On Mar 1, 2024, at 11:37 PM, Zhou, Hui <zhouh(a)anl.gov> wrote: I don't know what is causing the error, but here are a few work arounds: 1. You may try configure with --with-device=ch3. It uses an older device that is in maintenance mode, but likely it will work for you. 2. Try install libfabric separately and configure mpich with --with-libfabric=prefix 3. You may try brew install libfabric or brew install mpich. -- Hui ________________________________ From: Douglas G Dommermuth via discuss <discuss(a)mpich.org<mailto:[email protected]>> Sent: Friday, March 1, 2024 6:09 PM To: discuss(a)mpich.org<mailto:[email protected]> <discuss(a)mpich.org<mailto:[email protected]>> Cc: Douglas G Dommermuth <dgd(a)mit.edu<mailto:[email protected]>> Subject: [mpich-discuss] Install errors on MAC OS Sonoma: "error: expected declaration specifiers or ‘...’ before string constant" This Message Is From an External Sender This message came from outside your organization. I am compiling mpich 4.2.0 on a MacBook Air Apple M1 running macOS Sonoma 14.3.1. I am using gcc and gfortran versions 13.2.0. I get the following errors when I build MPICH: /Applications/Xcode.app/Contents/Developer/usr/bin/make all-am depbase=`echo util/info.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\ gcc -DHAVE_CONFIG_H -I. -I./include -D_GNU_SOURCE -D__USE_XOPEN2K8 -DSYSCONFDIR=\"/Users/douglasdommermuth/mpich2/mpich2-install/etc\" -DRDMADIR=\"@rdmadir@\" -DPROVDLDIR=\"/Users/douglasdommermuth/mpich2/mpich2-install/lib/libfabric\" -I./prov/sockets/include -I./prov/sockets -Wall -O2 -DNDEBUG -MT util/info.o -MD -MP -MF $depbase.Tpo -c -o util/info.o util/info.c &&\ mv -f $depbase.Tpo $depbase.Po In file included from ./include/osx/osd.h:182, from ./include/ofi_osd.h:60, from util/info.c:38: /Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/os/lock.h:114:33: error: expected declaration specifiers or ‘...’ before string constant 114 | OS_SWIFT_UNAVAILABLE_FROM_ASYNC("Use OSAllocatedUnfairLock.performWhileLocked() for async-safe scoped locking") | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ /Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/os/lock.h:140:33: error: expected declaration specifiers or ‘...’ before string constant 140 | OS_SWIFT_UNAVAILABLE_FROM_ASYNC("Use OSAllocatedUnfairLock.tryPerformWhileLocked() for async-safe scoped locking") | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ /Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/os/lock.h:154:33: error: expected declaration specifiers or ‘...’ before string constant 154 | OS_SWIFT_UNAVAILABLE_FROM_ASYNC("Use OSAllocatedUnfairLock.performWhileLocked() for async-safe scoped locking") | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ./include/osx/osd.h: In function ‘pthread_spin_lock’: ./include/osx/osd.h:194:9: warning: implicit declaration of function ‘os_unfair_lock_lock’ [-Wimplicit-function-declaration] 194 | os_unfair_lock_lock(lock); | ^~~~~~~~~~~~~~~~~~~ ./include/osx/osd.h: In function ‘pthread_spin_unlock’: ./include/osx/osd.h:200:9: warning: implicit declaration of function ‘os_unfair_lock_unlock’ [-Wimplicit-function-declaration] 200 | os_unfair_lock_unlock(lock); | ^~~~~~~~~~~~~~~~~~~~~ ./include/osx/osd.h: In function ‘pthread_spin_trylock’: ./include/osx/osd.h:206:16: warning: implicit declaration of function ‘os_unfair_lock_trylock’ [-Wimplicit-function-declaration] 206 | return os_unfair_lock_trylock(lock) ? 0 : EBUSY; | ^~~~~~~~~~~~~~~~~~~~~~ make[3]: *** [util/info.o] Error 1 make[2]: *** [all] Error 2 make[1]: *** [all-recursive] Error 1 make: *** [all] Error 2 It looks like errors in macOS’s command line tools. I don’t know how to circumvent the problem. Various text and log files are attached in mpich2.tar.gz. A copy of the lock.h code is also included in the zip file. Thank you, Doug.

1 0