- discuss - mailman3.cels.anl.gov

Re: [mpich-discuss] Running configure on Suse Enterprise 10
by anton.s.murfin＠sellafieldsites.com 05 Jul '13

05 Jul '13

1 0

Re: [mpich-discuss] Running configure on Suse Enterprise 10
by William Gropp 05 Jul '13

05 Jul '13

Can you also send that config.log file that was created - this contains more details; in particular, I need to see why configure thought that the Fortran compiler worked but then was unable to use it to determine properties such as the size of a Fortran INTEGER. Bill William Gropp Director, Parallel Computing Institute Deputy Director for Research Institute for Advanced Computing Applications and Technologies Thomas M. Siebel Chair in Computer Science University of Illinois Urbana-Champaign On Jul 5, 2013, at 6:15 AM, <anton.s.murfin(a)sellafieldsites.com> wrote: > > Dear MPICH developers, > > I have downloaded and extracted mpich-3.0.4.tar.gz to /home/ris3/asm7/mpich-3.0.4/ > > I then created a build directory as /home/ris3/asm7/mpich-3.0.4_build/ > > I went back to /home/ris3/asm7/mpich-3.0.4/ and executed the .configure script detailed below > > ./configure --prefix = /home/ris3/asm7/mpich-3.0.4_build 2>&1 | tee c.txt (I confirmed I was using bash shell, using echo $SHELL) > > The output from this c.txt is attached and there is only one error right at the bottom although there are some warnings. > > No Makefile is created so the next step make doesn't work. > > I just wanted to see if you can see what the issue is. > > Many thanks > > Anton > > > "The information contained in this email may be commercially sensitive and/or legally privileged. It is intended solely for the person(s) to whom it is addressed. If you are not a named recipient, you are on notice of its status. Please notify the sender immediately by reply e-mail and then delete this message from your system. You must not disclose it to any other person, copy or distribute it or use it for any purpose. > > Views expressed in this email are not necessarily those of Sellafield Ltd. > > Sellafield Ltd, a company owned by Nuclear Management Partners Ltd, is registered in England and Wales, Company number 1002607. The registered office is situated at Booths Park, Chelford Road, Knutsford, Cheshire WA16 8QZ." > > <c.txt>_______________________________________________ > 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] Running configure on Suse Enterprise 10
by Jeff Hammond 05 Jul '13

05 Jul '13

Disable Fortran if you don't have a Fortran compiler. The options are shown with the help invocation. Jeff Sent from my iPhone On Jul 5, 2013, at 6:15 AM, "anton.s.murfin(a)sellafieldsites.com" < anton.s.murfin(a)sellafieldsites.com> wrote: Dear MPICH developers, I have downloaded and extracted mpich-3.0.4.tar.gz to /home/ris3/asm7/mpich-3.0.4/ I then created a build directory as /home/ris3/asm7/mpich-3.0.4_build/ I went back to /home/ris3/asm7/mpich-3.0.4/ and executed the .configure script detailed below *./configure --prefix = /home/ris3/asm7/mpich-3.0.4_build 2>&1 | tee c.txt* (I confirmed I was using bash shell, using echo $SHELL) The output from this c.txt is attached and there is only one error right at the bottom although there are some warnings. No Makefile is created so the next step make doesn't work. I just wanted to see if you can see what the issue is. Many thanks Anton "The information contained in this email may be commercially sensitive and/or legally privileged. It is intended solely for the person(s) to whom it is addressed. If you are not a named recipient, you are on notice of its status. Please notify the sender immediately by reply e-mail and then delete this message from your system. You must not disclose it to any other person, copy or distribute it or use it for any purpose. Views expressed in this email are not necessarily those of Sellafield Ltd. Sellafield Ltd, a company owned by Nuclear Management Partners Ltd, is registered in England and Wales, Company number 1002607. The registered office is situated at Booths Park, Chelford Road, Knutsford, Cheshire WA16 8QZ." <c.txt> _______________________________________________ discuss mailing list discuss(a)mpich.org To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss

1 0

Running configure on Suse Enterprise 10
by anton.s.murfin＠sellafieldsites.com 05 Jul '13

05 Jul '13

1 0

Re: [mpich-discuss] beginner for remote mem access
by Jeff Hammond 03 Jul '13

03 Jul '13

1) When you reply to the digest emails, change the subject to that of the thread you care to respond to. That will make it much easier for others to track the thread. 2) MPI_Win_fence is not similar to MPI_Barrier in some cases. The MPICH implementation can turn some calls into no-ops and others into MPI_Reduce_scatter. 3) Regarding "use MPI window as thread level", I really don't understand your thinking at all. MPI RMA is not shared memory nor is MPI a threading model. Jeff On Wed, Jul 3, 2013 at 2:46 PM, Sufeng Niu <sniu(a)hawk.iit.edu> wrote: > > Thanks a lot for your guys reply. I should initial b array, I figure out the > MPI_Win_fence should be called by all processes, which will solve the > assertion failed issue. > > it seems that MPI_Win_fence is very similar to MPI_Barrier, Can I say > MPI_Win_fence is the RMA version of MPI_Barrier? > > sorry about so many questions. Can I use MPI window as thread level? let's > say I have 4 processes, each process has 8 threads. if only one thread is > used to create mem access window, other process copy it and internal 8 > threads use shared memory to perform some operations. Is that ok? > > Thanks a lot! > > > > On Wed, Jul 3, 2013 at 1:28 PM, <discuss-request(a)mpich.org> wrote: >> >> Send discuss mailing list submissions to >> discuss(a)mpich.org >> >> To subscribe or unsubscribe via the World Wide Web, visit >> https://lists.mpich.org/mailman/listinfo/discuss >> or, via email, send a message with subject or body 'help' to >> discuss-request(a)mpich.org >> >> You can reach the person managing the list at >> discuss-owner(a)mpich.org >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of discuss digest..." >> >> >> Today's Topics: >> >> 1. beginner for remote mem access (Sufeng Niu) >> 2. Re: beginner for remote mem access (Yi Gu) >> 3. Re: beginner for remote mem access (Yi Gu) >> 4. Re: beginner for remote mem access (Jeff Hammond) >> >> >> ---------------------------------------------------------------------- >> >> Message: 1 >> Date: Wed, 3 Jul 2013 13:12:46 -0500 >> From: Sufeng Niu <sniu(a)hawk.iit.edu> >> To: discuss(a)mpich.org >> Subject: [mpich-discuss] beginner for remote mem access >> Message-ID: >> >> <CAFNNHkwxqXUB_+b8_tZ=L5K7VcDVhsz4ZMDDMGF55v-FAhgeqg(a)mail.gmail.com> >> Content-Type: text/plain; charset="iso-8859-1" >> >> Hi, >> >> I am a beginner and just try to use remote memory access, and I wrote a >> simple program to test it: >> >> #include "mpi.h" >> #include <stdio.h> >> #define SIZE 8 >> >> int main(int argc, char *argv[]) >> { >> int numtasks, rank, source=0, dest, tag=1, i; >> float a[64] = >> { >> 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, >> 12.0, 13.0, 14.0, 15.0, 16.0, >> 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, >> 12.0, 13.0, 14.0, 15.0, 16.0, >> 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, >> 12.0, 13.0, 14.0, 15.0, 16.0, >> 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, >> 12.0, 13.0, 14.0, 15.0, 16.0, >> }; >> float b[SIZE]; >> >> MPI_Status stat; >> >> MPI_Init(&argc,&argv); >> MPI_Comm_rank(MPI_COMM_WORLD, &rank); >> MPI_Comm_size(MPI_COMM_WORLD, &numtasks); >> >> MPI_Win win; >> >> // check processor rank >> char processor_name[MPI_MAX_PROCESSOR_NAME]; >> int name_len; >> MPI_Get_processor_name(processor_name, &name_len); >> printf("-- processor %s, rank %d out of %d processors\n", >> processor_name, rank, numtasks); >> >> MPI_Barrier(MPI_COMM_WORLD); >> >> if (numtasks == 4) { >> if (rank == 0) { >> printf("create window \n"); >> MPI_Win_create(a, 8*sizeof(float), sizeof(float), >> MPI_INFO_NULL, MPI_COMM_WORLD, &win); >> >> } >> else { >> MPI_Win_create(MPI_BOTTOM, 0, sizeof(float), >> MPI_INFO_NULL, MPI_COMM_WORLD, &win); >> } >> >> MPI_Win_fence(0, win); >> >> if (rank == 1){ >> MPI_Get(b, SIZE, MPI_FLOAT, 0, 8, SIZE, MPI_FLOAT, >> win); >> >> MPI_Win_fence(0, win); >> } >> >> printf("rank= %d b= %3.1f %3.1f %3.1f %3.1f %3.1f %3.1f >> %3.1f %3.1f\n", rank,b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7]); >> } >> else >> printf("Must specify %d processors. Terminating.\n",SIZE); >> >> MPI_Win_free(&win); >> MPI_Finalize(); >> } >> >> However the terminal gives some odd results: >> rank= 0 b= 0.0 0.0 0.0 0.0 0.0 0.0 -71847793475452928.0 0.0 >> rank= 2 b= 0.0 0.0 0.0 0.0 0.0 0.0 222086852849451401216.0 0.0 >> rank= 3 b= 0.0 0.0 0.0 0.0 0.0 0.0 -74882.4 0.0 >> rank= 1 b= 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 >> the rank 1 is correct results, but others should be all zero. >> terminal also give some comments: "Assertion failed in file >> src/mpid/ch3/src/ch3u_rma_sync.c at line 5061: win_ptr->my_counter >= 0 >> internal ABORT - process 0" >> >> another question is if I use remote memory access. all process which does >> not create window for share must add additional line: >> MPI_Win_create(MPI_BOTTOM, 0, data_type, MPI_INFO_NULL, MPI_COMM_WORLD, >> &win); correct? >> >> Thanks a lot! >> >> -- >> Best Regards, >> Sufeng Niu >> ECASP lab, ECE department, Illinois Institute of Technology >> Tel: 312-731-7219 >>

1 0

Re: [mpich-discuss] discuss Digest, Vol 9, Issue 5
by Sufeng Niu 03 Jul '13

03 Jul '13

Thanks a lot for your guys reply. I should initial b array, I figure out the MPI_Win_fence should be called by all processes, which will solve the assertion failed issue. it seems that MPI_Win_fence is very similar to MPI_Barrier, Can I say MPI_Win_fence is the RMA version of MPI_Barrier? sorry about so many questions. Can I use MPI window as thread level? let's say I have 4 processes, each process has 8 threads. if only one thread is used to create mem access window, other process copy it and internal 8 threads use shared memory to perform some operations. Is that ok? Thanks a lot! On Wed, Jul 3, 2013 at 1:28 PM, <discuss-request(a)mpich.org> wrote: > Send discuss mailing list submissions to > discuss(a)mpich.org > > To subscribe or unsubscribe via the World Wide Web, visit > https://lists.mpich.org/mailman/listinfo/discuss > or, via email, send a message with subject or body 'help' to > discuss-request(a)mpich.org > > You can reach the person managing the list at > discuss-owner(a)mpich.org > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of discuss digest..." > > > Today's Topics: > > 1. beginner for remote mem access (Sufeng Niu) > 2. Re: beginner for remote mem access (Yi Gu) > 3. Re: beginner for remote mem access (Yi Gu) > 4. Re: beginner for remote mem access (Jeff Hammond) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Wed, 3 Jul 2013 13:12:46 -0500 > From: Sufeng Niu <sniu(a)hawk.iit.edu> > To: discuss(a)mpich.org > Subject: [mpich-discuss] beginner for remote mem access > Message-ID: > <CAFNNHkwxqXUB_+b8_tZ= > L5K7VcDVhsz4ZMDDMGF55v-FAhgeqg(a)mail.gmail.com> > Content-Type: text/plain; charset="iso-8859-1" > > Hi, > > I am a beginner and just try to use remote memory access, and I wrote a > simple program to test it: > > #include "mpi.h" > #include <stdio.h> > #define SIZE 8 > > int main(int argc, char *argv[]) > { > int numtasks, rank, source=0, dest, tag=1, i; > float a[64] = > { > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > }; > float b[SIZE]; > > MPI_Status stat; > > MPI_Init(&argc,&argv); > MPI_Comm_rank(MPI_COMM_WORLD, &rank); > MPI_Comm_size(MPI_COMM_WORLD, &numtasks); > > MPI_Win win; > > // check processor rank > char processor_name[MPI_MAX_PROCESSOR_NAME]; > int name_len; > MPI_Get_processor_name(processor_name, &name_len); > printf("-- processor %s, rank %d out of %d processors\n", > processor_name, rank, numtasks); > > MPI_Barrier(MPI_COMM_WORLD); > > if (numtasks == 4) { > if (rank == 0) { > printf("create window \n"); > MPI_Win_create(a, 8*sizeof(float), sizeof(float), > MPI_INFO_NULL, MPI_COMM_WORLD, &win); > > } > else { > MPI_Win_create(MPI_BOTTOM, 0, sizeof(float), > MPI_INFO_NULL, MPI_COMM_WORLD, &win); > } > > MPI_Win_fence(0, win); > > if (rank == 1){ > MPI_Get(b, SIZE, MPI_FLOAT, 0, 8, SIZE, MPI_FLOAT, > win); > > MPI_Win_fence(0, win); > } > > printf("rank= %d b= %3.1f %3.1f %3.1f %3.1f %3.1f %3.1f > %3.1f %3.1f\n", rank,b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7]); > } > else > printf("Must specify %d processors. Terminating.\n",SIZE); > > MPI_Win_free(&win); > MPI_Finalize(); > } > > However the terminal gives some odd results: > rank= 0 b= 0.0 0.0 0.0 0.0 0.0 0.0 -71847793475452928.0 0.0 > rank= 2 b= 0.0 0.0 0.0 0.0 0.0 0.0 222086852849451401216.0 0.0 > rank= 3 b= 0.0 0.0 0.0 0.0 0.0 0.0 -74882.4 0.0 > rank= 1 b= 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 > the rank 1 is correct results, but others should be all zero. > terminal also give some comments: "Assertion failed in file > src/mpid/ch3/src/ch3u_rma_sync.c at line 5061: win_ptr->my_counter >= 0 > internal ABORT - process 0" > > another question is if I use remote memory access. all process which does > not create window for share must add additional line: > MPI_Win_create(MPI_BOTTOM, 0, data_type, MPI_INFO_NULL, MPI_COMM_WORLD, > &win); correct? > > Thanks a lot! > > -- > Best Regards, > Sufeng Niu > ECASP lab, ECE department, Illinois Institute of Technology > Tel: 312-731-7219 >

1 0

Re: [mpich-discuss] beginner for remote mem access
by Jeff Hammond 03 Jul '13

03 Jul '13

Look at the MPICH test suite for numerous examples of correct RMA programs. Jeff On Wed, Jul 3, 2013 at 1:12 PM, Sufeng Niu <sniu(a)hawk.iit.edu> wrote: > Hi, > > I am a beginner and just try to use remote memory access, and I wrote a > simple program to test it: > > #include "mpi.h" > #include <stdio.h> > #define SIZE 8 > > int main(int argc, char *argv[]) > { > int numtasks, rank, source=0, dest, tag=1, i; > float a[64] = > { > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > }; > float b[SIZE]; > > MPI_Status stat; > > MPI_Init(&argc,&argv); > MPI_Comm_rank(MPI_COMM_WORLD, &rank); > MPI_Comm_size(MPI_COMM_WORLD, &numtasks); > > MPI_Win win; > > // check processor rank > char processor_name[MPI_MAX_PROCESSOR_NAME]; > int name_len; > MPI_Get_processor_name(processor_name, &name_len); > printf("-- processor %s, rank %d out of %d processors\n", > processor_name, rank, numtasks); > > MPI_Barrier(MPI_COMM_WORLD); > > if (numtasks == 4) { > if (rank == 0) { > printf("create window \n"); > MPI_Win_create(a, 8*sizeof(float), sizeof(float), > MPI_INFO_NULL, MPI_COMM_WORLD, &win); > > } > else { > MPI_Win_create(MPI_BOTTOM, 0, sizeof(float), > MPI_INFO_NULL, MPI_COMM_WORLD, &win); > } > > MPI_Win_fence(0, win); > > if (rank == 1){ > MPI_Get(b, SIZE, MPI_FLOAT, 0, 8, SIZE, MPI_FLOAT, > win); > > MPI_Win_fence(0, win); > } > > printf("rank= %d b= %3.1f %3.1f %3.1f %3.1f %3.1f %3.1f > %3.1f %3.1f\n", rank,b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7]); > } > else > printf("Must specify %d processors. Terminating.\n",SIZE); > > MPI_Win_free(&win); > MPI_Finalize(); > } > > However the terminal gives some odd results: > rank= 0 b= 0.0 0.0 0.0 0.0 0.0 0.0 -71847793475452928.0 0.0 > rank= 2 b= 0.0 0.0 0.0 0.0 0.0 0.0 222086852849451401216.0 0.0 > rank= 3 b= 0.0 0.0 0.0 0.0 0.0 0.0 -74882.4 0.0 > rank= 1 b= 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 > the rank 1 is correct results, but others should be all zero. > terminal also give some comments: "Assertion failed in file > src/mpid/ch3/src/ch3u_rma_sync.c at line 5061: win_ptr->my_counter >= 0 > internal ABORT - process 0" > > another question is if I use remote memory access. all process which does > not create window for share must add additional line: > MPI_Win_create(MPI_BOTTOM, 0, data_type, MPI_INFO_NULL, MPI_COMM_WORLD, > &win); correct? > > Thanks a lot! > > -- > 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 -- Jeff Hammond jeff.science(a)gmail.com

1 0

Re: [mpich-discuss] beginner for remote mem access
by Yi Gu 03 Jul '13

03 Jul '13

Hi, sufeng: I am not sure the second problem is, but I think MPI_Win_create(MPI_BOTTOM, 0, sizeof(float), MPI_INFO_NULL, MPI_COMM_WORLD, &win); should be MPI_Win_create(MPI_BOTTOM, 0, 1, MPI_INFO_NULL, MPI_COMM_WORLD, &win); For the third question, if they need to access, yes. Yi On Wed, Jul 3, 2013 at 1:18 PM, Yi Gu <gyi(a)mtu.edu> wrote: > Hi, sufeng: > > I think you may first initialize array first, since you print out b > without initialization, > it could print out anything. > > Yi > > > On Wed, Jul 3, 2013 at 1:12 PM, Sufeng Niu <sniu(a)hawk.iit.edu> wrote: > >> Hi, >> >> I am a beginner and just try to use remote memory access, and I wrote a >> simple program to test it: >> >> #include "mpi.h" >> #include <stdio.h> >> #define SIZE 8 >> >> int main(int argc, char *argv[]) >> { >> int numtasks, rank, source=0, dest, tag=1, i; >> float a[64] = >> { >> 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, >> 12.0, 13.0, 14.0, 15.0, 16.0, >> 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, >> 12.0, 13.0, 14.0, 15.0, 16.0, >> 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, >> 12.0, 13.0, 14.0, 15.0, 16.0, >> 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, >> 12.0, 13.0, 14.0, 15.0, 16.0, >> }; >> float b[SIZE]; >> >> MPI_Status stat; >> >> MPI_Init(&argc,&argv); >> MPI_Comm_rank(MPI_COMM_WORLD, &rank); >> MPI_Comm_size(MPI_COMM_WORLD, &numtasks); >> >> MPI_Win win; >> >> // check processor rank >> char processor_name[MPI_MAX_PROCESSOR_NAME]; >> int name_len; >> MPI_Get_processor_name(processor_name, &name_len); >> printf("-- processor %s, rank %d out of %d processors\n", >> processor_name, rank, numtasks); >> >> MPI_Barrier(MPI_COMM_WORLD); >> >> if (numtasks == 4) { >> if (rank == 0) { >> printf("create window \n"); >> MPI_Win_create(a, 8*sizeof(float), sizeof(float), >> MPI_INFO_NULL, MPI_COMM_WORLD, &win); >> >> } >> else { >> MPI_Win_create(MPI_BOTTOM, 0, sizeof(float), >> MPI_INFO_NULL, MPI_COMM_WORLD, &win); >> } >> >> MPI_Win_fence(0, win); >> >> if (rank == 1){ >> MPI_Get(b, SIZE, MPI_FLOAT, 0, 8, SIZE, >> MPI_FLOAT, win); >> >> MPI_Win_fence(0, win); >> } >> >> printf("rank= %d b= %3.1f %3.1f %3.1f %3.1f %3.1f %3.1f >> %3.1f %3.1f\n", rank,b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7]); >> } >> else >> printf("Must specify %d processors. Terminating.\n",SIZE); >> >> MPI_Win_free(&win); >> MPI_Finalize(); >> } >> >> However the terminal gives some odd results: >> rank= 0 b= 0.0 0.0 0.0 0.0 0.0 0.0 -71847793475452928.0 0.0 >> rank= 2 b= 0.0 0.0 0.0 0.0 0.0 0.0 222086852849451401216.0 0.0 >> rank= 3 b= 0.0 0.0 0.0 0.0 0.0 0.0 -74882.4 0.0 >> rank= 1 b= 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 >> the rank 1 is correct results, but others should be all zero. >> terminal also give some comments: "Assertion failed in file >> src/mpid/ch3/src/ch3u_rma_sync.c at line 5061: win_ptr->my_counter >= 0 >> internal ABORT - process 0" >> >> another question is if I use remote memory access. all process which does >> not create window for share must add additional line: >> MPI_Win_create(MPI_BOTTOM, 0, data_type, MPI_INFO_NULL, MPI_COMM_WORLD, >> &win); correct? >> >> Thanks a lot! >> >> -- >> 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 >> > >

1 0

Re: [mpich-discuss] beginner for remote mem access
by Yi Gu 03 Jul '13

03 Jul '13

Hi, sufeng: I think you may first initialize array first, since you print out b without initialization, it could print out anything. Yi On Wed, Jul 3, 2013 at 1:12 PM, Sufeng Niu <sniu(a)hawk.iit.edu> wrote: > Hi, > > I am a beginner and just try to use remote memory access, and I wrote a > simple program to test it: > > #include "mpi.h" > #include <stdio.h> > #define SIZE 8 > > int main(int argc, char *argv[]) > { > int numtasks, rank, source=0, dest, tag=1, i; > float a[64] = > { > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, > 12.0, 13.0, 14.0, 15.0, 16.0, > }; > float b[SIZE]; > > MPI_Status stat; > > MPI_Init(&argc,&argv); > MPI_Comm_rank(MPI_COMM_WORLD, &rank); > MPI_Comm_size(MPI_COMM_WORLD, &numtasks); > > MPI_Win win; > > // check processor rank > char processor_name[MPI_MAX_PROCESSOR_NAME]; > int name_len; > MPI_Get_processor_name(processor_name, &name_len); > printf("-- processor %s, rank %d out of %d processors\n", > processor_name, rank, numtasks); > > MPI_Barrier(MPI_COMM_WORLD); > > if (numtasks == 4) { > if (rank == 0) { > printf("create window \n"); > MPI_Win_create(a, 8*sizeof(float), sizeof(float), > MPI_INFO_NULL, MPI_COMM_WORLD, &win); > > } > else { > MPI_Win_create(MPI_BOTTOM, 0, sizeof(float), > MPI_INFO_NULL, MPI_COMM_WORLD, &win); > } > > MPI_Win_fence(0, win); > > if (rank == 1){ > MPI_Get(b, SIZE, MPI_FLOAT, 0, 8, SIZE, MPI_FLOAT, > win); > > MPI_Win_fence(0, win); > } > > printf("rank= %d b= %3.1f %3.1f %3.1f %3.1f %3.1f %3.1f > %3.1f %3.1f\n", rank,b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7]); > } > else > printf("Must specify %d processors. Terminating.\n",SIZE); > > MPI_Win_free(&win); > MPI_Finalize(); > } > > However the terminal gives some odd results: > rank= 0 b= 0.0 0.0 0.0 0.0 0.0 0.0 -71847793475452928.0 0.0 > rank= 2 b= 0.0 0.0 0.0 0.0 0.0 0.0 222086852849451401216.0 0.0 > rank= 3 b= 0.0 0.0 0.0 0.0 0.0 0.0 -74882.4 0.0 > rank= 1 b= 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 > the rank 1 is correct results, but others should be all zero. > terminal also give some comments: "Assertion failed in file > src/mpid/ch3/src/ch3u_rma_sync.c at line 5061: win_ptr->my_counter >= 0 > internal ABORT - process 0" > > another question is if I use remote memory access. all process which does > not create window for share must add additional line: > MPI_Win_create(MPI_BOTTOM, 0, data_type, MPI_INFO_NULL, MPI_COMM_WORLD, > &win); correct? > > Thanks a lot! > > -- > 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 >

1 0

beginner for remote mem access
by Sufeng Niu 03 Jul '13

03 Jul '13

Hi, I am a beginner and just try to use remote memory access, and I wrote a simple program to test it: #include "mpi.h" #include <stdio.h> #define SIZE 8 int main(int argc, char *argv[]) { int numtasks, rank, source=0, dest, tag=1, i; float a[64] = { 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0, 15.0, 16.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0, 15.0, 16.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0, 15.0, 16.0, 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0, 15.0, 16.0, }; float b[SIZE]; MPI_Status stat; MPI_Init(&argc,&argv); MPI_Comm_rank(MPI_COMM_WORLD, &rank); MPI_Comm_size(MPI_COMM_WORLD, &numtasks); MPI_Win win; // check processor rank char processor_name[MPI_MAX_PROCESSOR_NAME]; int name_len; MPI_Get_processor_name(processor_name, &name_len); printf("-- processor %s, rank %d out of %d processors\n", processor_name, rank, numtasks); MPI_Barrier(MPI_COMM_WORLD); if (numtasks == 4) { if (rank == 0) { printf("create window \n"); MPI_Win_create(a, 8*sizeof(float), sizeof(float), MPI_INFO_NULL, MPI_COMM_WORLD, &win); } else { MPI_Win_create(MPI_BOTTOM, 0, sizeof(float), MPI_INFO_NULL, MPI_COMM_WORLD, &win); } MPI_Win_fence(0, win); if (rank == 1){ MPI_Get(b, SIZE, MPI_FLOAT, 0, 8, SIZE, MPI_FLOAT, win); MPI_Win_fence(0, win); } printf("rank= %d b= %3.1f %3.1f %3.1f %3.1f %3.1f %3.1f %3.1f %3.1f\n", rank,b[0],b[1],b[2],b[3],b[4],b[5],b[6],b[7]); } else printf("Must specify %d processors. Terminating.\n",SIZE); MPI_Win_free(&win); MPI_Finalize(); } However the terminal gives some odd results: rank= 0 b= 0.0 0.0 0.0 0.0 0.0 0.0 -71847793475452928.0 0.0 rank= 2 b= 0.0 0.0 0.0 0.0 0.0 0.0 222086852849451401216.0 0.0 rank= 3 b= 0.0 0.0 0.0 0.0 0.0 0.0 -74882.4 0.0 rank= 1 b= 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 the rank 1 is correct results, but others should be all zero. terminal also give some comments: "Assertion failed in file src/mpid/ch3/src/ch3u_rma_sync.c at line 5061: win_ptr->my_counter >= 0 internal ABORT - process 0" another question is if I use remote memory access. all process which does not create window for share must add additional line: MPI_Win_create(MPI_BOTTOM, 0, data_type, MPI_INFO_NULL, MPI_COMM_WORLD, &win); correct? Thanks a lot! -- Best Regards, Sufeng Niu ECASP lab, ECE department, Illinois Institute of Technology Tel: 312-731-7219

1 0