From panda at cse.ohio-state.edu Fri May 1 00:29:41 2009 From: panda at cse.ohio-state.edu (Dhabaleswar Panda) Date: Fri May 1 00:29:57 2009 Subject: [mvapich-discuss] examples_collchk issue In-Reply-To: <49F9B0D1.8060902@mail.rb.ru> Message-ID: Thanks for your report. This example file and the associated directory are inherited from MPICH2. We have informed the MPICH2 team about this and they are taking a look at it. As soon as this is resolved by the MPICH2 team, we will reflect this in MVAPICH2 distribution. Thanks, DK On Thu, 30 Apr 2009, Arthur Yuldashev wrote: > *This message was transferred with a trial version of CommuniGate(r) Pro* > > Hello! > > It seems that we've found a bug in some of examples provided with > mvapich2-1.2. > For instance in time_alltoallv.c there are following strings of code: > > if ( argv != NULL && argv[1] != NULL ) > block_size = atoi( argv[1] ); > else > block_size = 1; > > if ( argv != NULL && argv[2] != NULL ) > num_itr = atoi( argv[2] ); > else > num_itr = 1; > > > We ran it without any command line arguments resulting in argv[1] equal > to NULL, > but argv[2] was equal to one of environment variables. > So atoi(argv[2]) was equal to 0 resulting in 0 number of iterations, > not 1 as supposed. > And so actually no MPI_Alltoallv communications were done. > > Best regards, > Arthur Yuldashev > > > _______________________________________________ > mvapich-discuss mailing list > mvapich-discuss@cse.ohio-state.edu > http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss > From kholcomb at virginia.edu Fri May 1 10:47:19 2009 From: kholcomb at virginia.edu (Katherine Holcomb) Date: Fri May 1 10:47:34 2009 Subject: [mvapich-discuss] Problems with RDMA Message-ID: <1241189239.3882.563.camel@crestone.itc.virginia.edu> We have a new cluster with Qlogics IB. Basic diagnostics provided by the vendor work fine, but when I try to run MPI using MVAPICH (1 or 2) I get [vbuf.c 230] Cannot register vbuf region Internal Error: invalid error code ffffffff (Ring Index out of range) in MPIDI_C H3I_RDMA_init:505 Fatal error in MPI_Init: Other MPI error, error stack: MPIR_Init_thread(310)....: Initialization failed MPID_Init(113)...........: channel initialization failed MPIDI_CH3_Init(168)......: MPIDI_CH3I_RDMA_init(505): [vbuf.c 230] Cannot register vbuf region Internal Error: invalid error code ffffffff (Ring Index out of range) in MPIDI_C H3I_RDMA_init:505 repeated once for each process. We don't have any experience with IB so we are baffled, and I have not found much useful information on Google. Thanks for any help. -- Katherine Holcomb, Ph.D. kholcomb@virginia.edu UVACSE I046 BSEL, Clark Hall (434) 982-5948 University of Virginia 22904 From panda at cse.ohio-state.edu Fri May 1 11:14:25 2009 From: panda at cse.ohio-state.edu (Dhabaleswar Panda) Date: Fri May 1 11:14:39 2009 Subject: [mvapich-discuss] Problems with RDMA In-Reply-To: <1241189239.3882.563.camel@crestone.itc.virginia.edu> Message-ID: Thanks for your report. A couple of questions: - Did you build/install the MPI on these systems or QLogic did? - Are you able to run basic MPI benchmarks (such as OSU MPI benchmarks)? - When you are saying QLogic IB, are the adapters Mellanox-based or QLogic-based? These two adapters are very different and provide different interfaces to the MPI layer. For Mellanox-based adapters, OpenFabrics-Gen2 interface needs to be used for both MVAPICH 1 and MVAPICH2. MVAPICH 1 also has a separate MVAPICH-PSM interface for QLogic IB adapters supporting PSM. Similar interface is being designed for MVAPICH2 and will be available in the upcoming MVAPICH2 release. - The error report suggests that the application is not able to register buffers. Take a look at Section 9.3.4 of MVAPICH2 1.2 user guide for details. The URL is as follows: http://mvapich.cse.ohio-state.edu/support/user_guide_mvapich2-1.2.html#x1-580009.3.4 Thanks, DK On Fri, 1 May 2009, Katherine Holcomb wrote: > We have a new cluster with Qlogics IB. Basic diagnostics provided by > the vendor work fine, but when I try to run MPI using MVAPICH (1 or 2) I > get > > [vbuf.c 230] Cannot register vbuf region > Internal Error: invalid error code ffffffff (Ring Index out of range) in > MPIDI_C > H3I_RDMA_init:505 > Fatal error in MPI_Init: > Other MPI error, error stack: > MPIR_Init_thread(310)....: Initialization failed > MPID_Init(113)...........: channel initialization failed > MPIDI_CH3_Init(168)......: > MPIDI_CH3I_RDMA_init(505): > [vbuf.c 230] Cannot register vbuf region > Internal Error: invalid error code ffffffff (Ring Index out of range) in > MPIDI_C > H3I_RDMA_init:505 > > repeated once for each process. > > We don't have any experience with IB so we are baffled, and I have not > found much useful information on Google. > > Thanks for any help. > > -- > Katherine Holcomb, Ph.D. kholcomb@virginia.edu > UVACSE > I046 BSEL, Clark Hall (434) 982-5948 > University of Virginia 22904 > > _______________________________________________ > mvapich-discuss mailing list > mvapich-discuss@cse.ohio-state.edu > http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss > From maya.usatu at gmail.com Thu May 7 08:01:48 2009 From: maya.usatu at gmail.com (Maya Khaliullina) Date: Thu May 7 08:02:05 2009 Subject: [mvapich-discuss] problem with intra-node communications Message-ID: Hello, When we run any MPI program on 4, 6 or 8 processes using intra-node communications only, job hangs at end stage (we believe it occurs during MPI_Finalize). But if we use mvapich2-0.98 or Intel MPI 3.1.26 it works fine. Have any idea about this problem? We compiled mvapich2-1.2 with following parameters: ./configure --prefix=/gpfs/bos/mvapich2-1.2 --enable-romio --disable-debuginfo --enable-sharedlibs=gcc \ --enable-base-cache --with-rdma=gen2 --with-thread-package=pthreads CC=icc CFLAGS=-O3 Parameters of our HPC cluster: Node: 2xQuad Core Intel Xeon 2.33 GHz O/S: RHEL4.5 File System: GPFS Thanks, Maya -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20090507/efb4d4cb/attachment.html From perkinjo at cse.ohio-state.edu Thu May 7 08:21:41 2009 From: perkinjo at cse.ohio-state.edu (Jonathan Perkins) Date: Thu May 7 08:21:58 2009 Subject: [mvapich-discuss] problem with intra-node communications In-Reply-To: References: Message-ID: <20090507122141.GC3184@cse.ohio-state.edu> On Thu, May 07, 2009 at 06:01:48PM +0600, Maya Khaliullina wrote: > Hello, > When we run any MPI program on 4, 6 or 8 processes using intra-node > communications only, job hangs at end stage (we believe it occurs during > MPI_Finalize). But if we use mvapich2-0.98 or Intel MPI 3.1.26 it works > fine. > Have any idea about this problem? Have you tried mvapich2-1.2p1? This contains a patch that may resolve your issue. Please let us know if it does. > We compiled mvapich2-1.2 with following parameters: > ./configure --prefix=/gpfs/bos/mvapich2-1.2 --enable-romio > --disable-debuginfo --enable-sharedlibs=gcc \ > --enable-base-cache --with-rdma=gen2 --with-thread-package=pthreads > CC=icc > CFLAGS=-O3 > Parameters of our HPC cluster: > Node: 2xQuad Core Intel Xeon 2.33 GHz > O/S: RHEL4.5 > File System: GPFS > > Thanks, > Maya > _______________________________________________ > mvapich-discuss mailing list > mvapich-discuss@cse.ohio-state.edu > http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss -- Jonathan Perkins http://www.cse.ohio-state.edu/~perkinjo -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 197 bytes Desc: not available Url : http://mail.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20090507/2492e893/attachment.bin From christian.guggenberger at rzg.mpg.de Thu May 7 08:34:37 2009 From: christian.guggenberger at rzg.mpg.de (Christian Guggenberger) Date: Thu May 7 08:34:50 2009 Subject: [mvapich-discuss] problem with intra-node communications In-Reply-To: References: Message-ID: <20090507123437.GE14805@bonnie.rzg.mpg.de> On Thu, May 07, 2009 at 06:01:48PM +0600, Maya Khaliullina wrote: > Hello, > When we run any MPI program on 4, 6 or 8 processes using intra-node > communications only, job hangs at end stage (we believe it occurs during > MPI_Finalize). But if we use mvapich2-0.98 or Intel MPI 3.1.26 it works > fine. > Have any idea about this problem? Just curious, will your application finish properly if you disable the SRQ feature at runtime ? e.g. something like mpiexec -n 4 -env MV2_USE_SRQ 0 ./a.out I see a quite similar behaviour with MT25208 based SDR cards, but not (with the very same software environment) with MT25418 [ConnectX IB DDR, PCIe 2.0 2.5GT/s] based cards. cheers. - Christian From koop at cse.ohio-state.edu Thu May 7 12:43:52 2009 From: koop at cse.ohio-state.edu (Matthew Koop) Date: Thu May 7 12:44:09 2009 Subject: [mvapich-discuss] problem with intra-node communications In-Reply-To: Message-ID: Hi Maya, Sorry for the inconvenience. For now, running with MV2_USE_SRQ=0 when running an intra-node job should solve this issue. We already have a fix for this issue queued for the next release of MVAPICH2. Matt On Thu, 7 May 2009, Maya Khaliullina wrote: > Hello, > When we run any MPI program on 4, 6 or 8 processes using intra-node > communications only, job hangs at end stage (we believe it occurs during > MPI_Finalize). But if we use mvapich2-0.98 or Intel MPI 3.1.26 it works > fine. > Have any idea about this problem? > We compiled mvapich2-1.2 with following parameters: > ./configure --prefix=/gpfs/bos/mvapich2-1.2 --enable-romio > --disable-debuginfo --enable-sharedlibs=gcc \ > --enable-base-cache --with-rdma=gen2 --with-thread-package=pthreads > CC=icc > CFLAGS=-O3 > Parameters of our HPC cluster: > Node: 2xQuad Core Intel Xeon 2.33 GHz > O/S: RHEL4.5 > File System: GPFS > > Thanks, > Maya > From maya.usatu at gmail.com Thu May 7 12:59:02 2009 From: maya.usatu at gmail.com (Maya Khaliullina) Date: Thu May 7 12:59:20 2009 Subject: Fwd: [mvapich-discuss] problem with intra-node communications In-Reply-To: References: Message-ID: ---------- Forwarded message ---------- From: Maya Khaliullina Date: 2009/5/7 Subject: Re: [mvapich-discuss] problem with intra-node communications To: Matthew Koop Thanks to all. Disabling of SRQ feature is really solves our problem. Thanks again, Maya 2009/5/7 Matthew Koop Hi Maya, > > Sorry for the inconvenience. For now, running with MV2_USE_SRQ=0 when > running an intra-node job should solve this issue. We already have a fix > for this issue queued for the next release of MVAPICH2. > > Matt > > On Thu, 7 May 2009, Maya Khaliullina wrote: > > > Hello, > > When we run any MPI program on 4, 6 or 8 processes using intra-node > > communications only, job hangs at end stage (we believe it occurs during > > MPI_Finalize). But if we use mvapich2-0.98 or Intel MPI 3.1.26 it works > > fine. > > Have any idea about this problem? > > We compiled mvapich2-1.2 with following parameters: > > ./configure --prefix=/gpfs/bos/mvapich2-1.2 --enable-romio > > --disable-debuginfo --enable-sharedlibs=gcc \ > > --enable-base-cache --with-rdma=gen2 --with-thread-package=pthreads > > CC=icc > > CFLAGS=-O3 > > Parameters of our HPC cluster: > > Node: 2xQuad Core Intel Xeon 2.33 GHz > > O/S: RHEL4.5 > > File System: GPFS > > > > Thanks, > > Maya > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20090507/66428bcf/attachment-0001.html From Adam.Lundrigan at dfo-mpo.gc.ca Tue May 12 08:08:37 2009 From: Adam.Lundrigan at dfo-mpo.gc.ca (Lundrigan, Adam) Date: Tue May 12 08:57:07 2009 Subject: [mvapich-discuss] Compiling MVAPICH2 on CentOS with Sun Studio for Linux Message-ID: Skipped content of type multipart/alternative-------------- next part -------------- A non-text attachment was scrubbed... Name: config.log Type: application/octet-stream Size: 176073 bytes Desc: config.log Url : http://mail.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20090512/120be594/config-0001.obj From arthur at mail.rb.ru Tue May 19 10:14:17 2009 From: arthur at mail.rb.ru (Arthur Yuldashev) Date: Tue May 19 10:14:40 2009 Subject: [mvapich-discuss] message coalescing Message-ID: <4A12BEB9.3000603@mail.rb.ru> Hello! I've read your presentation: Reducing Connection Memory Requirements of MPI for InfiniBand Clusters: A Message Coalescing Approach. Could you please explain one moment. I see that in the examples provided in your presentation different benchmarks were started with 256 processes, 4 per node. So messages going from different processes say of node A to node B are coalesced, aren't they? Does it work when I have several applications communicating from one node with other nodes, could messages from different applications be coalesced? Best regards, Arthur From panda at cse.ohio-state.edu Tue May 19 23:46:33 2009 From: panda at cse.ohio-state.edu (Dhabaleswar Panda) Date: Tue May 19 23:46:55 2009 Subject: [mvapich-discuss] message coalescing In-Reply-To: <4A12BEB9.3000603@mail.rb.ru> Message-ID: > Hello! > > I've read your presentation: Reducing Connection Memory Requirements of > MPI for InfiniBand Clusters: A Message Coalescing Approach. > Could you please explain one moment. > I see that in the examples provided in your presentation different > benchmarks were started with 256 processes, 4 per node. > So messages going from different processes say of node A to node B are > coalesced, aren't they? No, Coalescing is done per QP-basis. We are talking about MPI library here. So, messages across a pair of processes (say process 1 on node A with process 1 on node B) can be coalesed. There is no aggregation of messages from different processes on node A to other node. > Does it work when I have several applications communicating from one > node with other nodes, > could messages from different applications be coalesced? No. Thanks, DK > Best regards, > Arthur > > _______________________________________________ > mvapich-discuss mailing list > mvapich-discuss@cse.ohio-state.edu > http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss > From shtsai at uga.edu Thu May 21 14:54:56 2009 From: shtsai at uga.edu (Shan-ho Tsai) Date: Thu May 21 14:55:13 2009 Subject: [mvapich-discuss] undefined reference to vbuf_fast_rdma_alloc and vbuf_rdma_address_send Message-ID: <20090521145456.CPU12304@punts5.cc.uga.edu> Hello all, I am trying to build MVAPICH 1.1 (2009.05.19) using PGI v.7.2 compilers, on an Intel quad-core Xeon, running Linux RHEL4, and Qlogic Infiniband (Mellanox adapter). I am using OpenFabrics/Gen2. I edited make.mvapich.gen2 to set the path and compilers to IBHOME=${IBHOME:-/usr} IBHOME_LIB=${IBHOME_LIB:-/usr/lib64} PREFIX=${PREFIX:-/usr/local/mvapich-1.1} export CC=${CC:-pgcc} export CXX=${CXX:-pgCC} export F77=${F77:-pgf77} export F90=${F90:-pgf90} and I added -DMEMORY_RELIABLE to CFLAGS. The configuration and compilation appear fine (no error messages), but I got an error in the test to see if MPI applications can be linked. The error is Testing that MPI applications can be linked... make --no-print-directory linktest make overtake /usr/local/src/infiniband/qlogic/mpi/mvapich-1.1-2009-05-19/bin/mpicc -D_EM64T_ -DEARLY_SEND_COMPLETION -DMEMORY_SCALE -DMEMORY_RELIABLE -DVIADEV_RPUT_SUPPORT -D_SMP_ -D_SMP_RNDV_ -DXRC -DCH_GEN2 -D_GNU_SOURCE -I/usr/include -O -DHAVE_MPICHCONF_H -DHAVE_STDLIB_H=1 -DHAVE_UNISTD_H=1 -DHAVE_STRING_H=1 -DUSE_STDARG=1 -DHAVE_LONG_DOUBLE=1 -DHAVE_LONG_LONG_INT=1 -DHAVE_PROTOTYPES=1 -DHAVE_SIGNAL_H=1 -DHAVE_SIGACTION=1 -DHAVE_SLEEP=1 -DHAVE_SYSCONF=1 -c overtake.c /usr/local/src/infiniband/qlogic/mpi/mvapich-1.1-2009-05-19/bin/mpicc -D_EM64T_ -DEARLY_SEND_COMPLETION -DMEMORY_SCALE -DMEMORY_RELIABLE -DVIADEV_RPUT_SUPPORT -D_SMP_ -D_SMP_RNDV_ -DXRC -DCH_GEN2 -D_GNU_SOURCE -I/usr/include -O -DHAVE_MPICHCONF_H -DHAVE_STDLIB_H=1 -DHAVE_UNISTD_H=1 -DHAVE_STRING_H=1 -DUSE_STDARG=1 -DHAVE_LONG_DOUBLE=1 -DHAVE_LONG_LONG_INT=1 -DHAVE_PROTOTYPES=1 -DHAVE_SIGNAL_H=1 -DHAVE_SIGACTION=1 -DHAVE_SLEEP=1 -DHAVE_SYSCONF=1 -c test.c /usr/local/src/infiniband/qlogic/mpi/mvapich-1.1-2009-05-19/bin/mpicc -o overtake overtake.o test.o /usr/local/src/infiniband/qlogic/mpi/mvapich-1.1-2009-05-19/lib/libmpich.a(cm_user.o)(.text+0x21a): In function `odu_enable_qp': : undefined reference to `vbuf_fast_rdma_alloc' /usr/local/src/infiniband/qlogic/mpi/mvapich-1.1-2009-05-19/lib/libmpich.a(cm_user.o)(.text+0x224): In function `odu_enable_qp': : undefined reference to `vbuf_rdma_address_send' make[4]: *** [overtake] Error 2 make[3]: [linktest] Error 2 (ignored) Could not link a C program with MPI libraries make[3]: *** [linktest] Error 1 make[2]: *** [linktest] Error 2 make[1]: *** [mpi-lib-test] Error 2 make: *** [mpi] Error 2 Failure in building MVAPICH. Are vbuf_rdma_address_send and vbuf_fast_rdma_alloc defined in an mpich library or in an ofed library? I am not sure what is causing this error. Should I have defined some other compilation macros? Any suggestions how I can get this problem resolved? Thank you very much in advance. Regards, Shan-Ho Tsai ----------------------------------------- Shan-Ho Tsai University of Georgia, Athens GA, USA From yentj at infowrap.com.tw Tue May 26 04:03:49 2009 From: yentj at infowrap.com.tw (Ting-jen Yen) Date: Tue May 26 04:04:55 2009 Subject: [mvapich-discuss] mvapich2-1.2p1 program hang Message-ID: <1243325029.4147.34.camel@yentj.infowrap.com.tw> I have some problem when running MPI programs using mvapich2-1.2p1. A program is running fine when it is using more than one nodes. However, if it is using only one node, (that is, 4 cpu cores or 8 cpu cores, for example,) it would sometimes stop at MPI_Finalize(), and hang there forever. I have tried various programs, even the "hello world" example, with similar results. (This does not always happen. Sometimes, the programs run and finish just fine.) The system is a cluster of more than 80 IBM blade, each with 2 quad core Xeon E5355 (that is, 8 CPU core per node). The OS is RHEL 4 update 4, and the InfiniBand driver used is OFED 1.3. The mvapich2 was compiled using Intel compiler, with no explicit RDMA option, which should mean "gen2" by default. Any idea what could cause this problem? Thanks, Ting-jen From koop at cse.ohio-state.edu Tue May 26 04:53:30 2009 From: koop at cse.ohio-state.edu (Matthew Koop) Date: Tue May 26 04:53:48 2009 Subject: [mvapich-discuss] mvapich2-1.2p1 program hang In-Reply-To: <1243325029.4147.34.camel@yentj.infowrap.com.tw> Message-ID: Ting-jen, Can you try added MV2_USE_SRQ=0 when running on a single node? If this works, then we already have a fix for this issue queued for the next release of MVAPICH2. Matt On Tue, 26 May 2009, Ting-jen Yen wrote: > > I have some problem when running MPI programs using mvapich2-1.2p1. > > A program is running fine when it is using more than one nodes. > However, if it is using only one node, (that is, 4 cpu cores or 8 cpu > cores, for example,) it would sometimes stop at MPI_Finalize(), and hang > there forever. I have tried various programs, even the "hello world" > example, with similar results. (This does not always happen. Sometimes, > the programs run and finish just fine.) > > The system is a cluster of more than 80 IBM blade, each with 2 quad > core Xeon E5355 (that is, 8 CPU core per node). The OS is RHEL 4 > update 4, and the InfiniBand driver used is OFED 1.3. > > The mvapich2 was compiled using Intel compiler, with no explicit RDMA > option, which should mean "gen2" by default. > > Any idea what could cause this problem? > > Thanks, > Ting-jen > > > _______________________________________________ > mvapich-discuss mailing list > mvapich-discuss@cse.ohio-state.edu > http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss > From yentj at infowrap.com.tw Tue May 26 05:01:14 2009 From: yentj at infowrap.com.tw (Ting-jen Yen) Date: Tue May 26 05:02:27 2009 Subject: [mvapich-discuss] mvapich2-1.2p1 program hang In-Reply-To: References: Message-ID: <1243328473.4147.38.camel@yentj.infowrap.com.tw> Hi, Matt, Thanks a lot, it is working now after I added this parameter. -- Ting-jen ¦b ¤G, 2009-05-26 16:53, Matthew Koop ¼g¹D¡G > Ting-jen, > > Can you try added MV2_USE_SRQ=0 when running on a single node? If this > works, then we already have a fix for this issue queued for the next > release of MVAPICH2. > > Matt > > > On Tue, 26 May 2009, Ting-jen Yen wrote: > > > > > I have some problem when running MPI programs using mvapich2-1.2p1. > > > > A program is running fine when it is using more than one nodes. > > However, if it is using only one node, (that is, 4 cpu cores or 8 cpu > > cores, for example,) it would sometimes stop at MPI_Finalize(), and hang > > there forever. I have tried various programs, even the "hello world" > > example, with similar results. (This does not always happen. Sometimes, > > the programs run and finish just fine.) > > > > The system is a cluster of more than 80 IBM blade, each with 2 quad > > core Xeon E5355 (that is, 8 CPU core per node). The OS is RHEL 4 > > update 4, and the InfiniBand driver used is OFED 1.3. > > > > The mvapich2 was compiled using Intel compiler, with no explicit RDMA > > option, which should mean "gen2" by default. > > > > Any idea what could cause this problem? > > > > Thanks, > > Ting-jen > > > > > > _______________________________________________ > > mvapich-discuss mailing list > > mvapich-discuss@cse.ohio-state.edu > > http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss > > > > From gagang at cdac.in Wed May 27 09:30:18 2009 From: gagang at cdac.in (Gagan Grover) Date: Wed May 27 15:56:09 2009 Subject: [mvapich-discuss] SRQ usage with UD QP Message-ID: Hi I am working on SRQ and want to test SRQ with UD as well as RC QP. I found the usage of SRQ with RC QP in Mvapich Gen2. But I want to know if there is any MPI application which also tests SRQ with UD QP. -- Gagan Grover Hardware Technology Development Group [HTDG], Centre for Development of Advanced Computing [C-DAC], Pune University Campus, Ganeshkhind, Pune. INDIA - 411007. Ph: +91 20 25704324. "A winner is not one who never fails, but one who NEVER QUITS!" From koop at cse.ohio-state.edu Thu May 28 05:12:43 2009 From: koop at cse.ohio-state.edu (Matthew Koop) Date: Thu May 28 05:13:02 2009 Subject: [mvapich-discuss] SRQ usage with UD QP In-Reply-To: Message-ID: None of the code in MVAPICH uses an SRQ for UD. There's much less of a reason to use an SRQ for UD since a single QP can communicate with any number of other QPs (and then just the RQ can be used). Matt On Wed, 27 May 2009, Gagan Grover wrote: > Hi > > I am working on SRQ and want to test SRQ with UD as well as RC QP. I found > the usage of SRQ with RC QP in Mvapich Gen2. But I want to know if there > is any MPI application which also tests SRQ with UD QP. > > > > -- > Gagan Grover > Hardware Technology Development Group [HTDG], > Centre for Development of Advanced Computing [C-DAC], > Pune University Campus, > Ganeshkhind, Pune. INDIA - 411007. > Ph: +91 20 25704324. > > "A winner is not one who never fails, but one who NEVER QUITS!" > > > _______________________________________________ > mvapich-discuss mailing list > mvapich-discuss@cse.ohio-state.edu > http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss > From federico.golfre at gmail.com Thu May 28 05:38:37 2009 From: federico.golfre at gmail.com (=?ISO-8859-1?Q?Federico_Golfr=E8_Andreasi?=) Date: Thu May 28 07:58:14 2009 Subject: [mvapich-discuss] MPI SPAWN with MVAPICH2 Message-ID: Hi! I have a small test that uses the MPI_COMM_SPAWN_MULTIPLE function. When I run it, it gives me this error: MPI_Comm_spawn_multiple(152)..: MPI_Comm_spawn_multiple(count=1, cmds=0x7fffae068470, argvs=0x7fffae068430, maxprocs=0x7fffae068450, infos=0x7fffae068410, root=0, MPI_COMM_WORLD, intercomm=0x7fffae068614, errors=0x7fffae0683f0) failed MPID_Comm_spawn_multiple(67)..: MPIDI_Comm_spawn_multiple(172): MPID_Open_port(69)............: Function not implemented It seems that that function is not implemented in MVAPICH-2, is it true? But MPI_Comm_spawn_multiple is a standard MPI2 function. I use mvapich2-1.2p1. Thank you! Federico -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20090528/dcef7cfa/attachment.html From panda at cse.ohio-state.edu Thu May 28 08:18:06 2009 From: panda at cse.ohio-state.edu (Dhabaleswar Panda) Date: Thu May 28 08:18:24 2009 Subject: [mvapich-discuss] MPI SPAWN with MVAPICH2 In-Reply-To: Message-ID: MVAPICH2 1.2p1 does not support this functionality. The upcoming version does. If you wait for a few days, you will be getting this feature. Thanks, DK > Hi! > > I have a small test that uses the MPI_COMM_SPAWN_MULTIPLE function. > > > When I run it, it gives me this error: > > MPI_Comm_spawn_multiple(152)..: MPI_Comm_spawn_multiple(count=1, > cmds=0x7fffae068470, argvs=0x7fffae068430, maxprocs=0x7fffae068450, > infos=0x7fffae068410, root=0, MPI_COMM_WORLD, intercomm=0x7fffae068614, > errors=0x7fffae0683f0) failed > MPID_Comm_spawn_multiple(67)..: > MPIDI_Comm_spawn_multiple(172): > MPID_Open_port(69)............: Function not implemented > > It seems that that function is not implemented in MVAPICH-2, is it true? > But MPI_Comm_spawn_multiple is a standard MPI2 function. > > > I use mvapich2-1.2p1. > > > > Thank you! > Federico >