Hi, Jeff

Thanks a lot for your reply. Sorry about subject issue. 

1.
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.
I am not quite sure about "turn some calls into no-ops and others into MPI_Reduce_scatter" Could you please give an example if possible? Another thing is if a process create a window for other process, but the data access is available after some operations. Should I use MPI_Win_fence or MPI_Barrier to sync? or other methods?

2.
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.
Sorry for uncleared statements, right now I would like to do multithreads and MPI hybrid programming. let me give an example: I have 3 processes, each one has 8 threads. thread 0 in process 0 creates a RMA window. if I would like all other threads to access it, should I use thread 0 in process 1, thread 0 in process 2 to MPI_Get the data from window, then use shared memory for internal threads to load the data? i am not sure what is the proper way for RMA in hybrid model.

Thank you so much!

Sufeng


On Fri, Jul 5, 2013 at 6:15 AM, <discuss-request@mpich.org> wrote:
Send discuss mailing list submissions to
        discuss@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@mpich.org

You can reach the person managing the list at
        discuss-owner@mpich.org

When replying, please edit your Subject line so it is more specific
than "Re: Contents of discuss digest..."


Today's Topics:

   1. Re:  beginner for remote mem access (Jeff Hammond)
   2.  Running configure on Suse Enterprise 10
      (anton.s.murfin@sellafieldsites.com)


----------------------------------------------------------------------

Message: 1
Date: Wed, 3 Jul 2013 14:50:44 -0500
From: Jeff Hammond <jeff.science@gmail.com>
To: discuss@mpich.org
Subject: Re: [mpich-discuss] beginner for remote mem access
Message-ID:
        <CAGKz=uJN2iH3Pxjxyq=YzrP=XG5Tp2+huxmBFTh+Lr7nVdLU4w@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1

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@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@mpich.org> wrote:
>>
>> Send discuss mailing list submissions to
>>         discuss@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@mpich.org
>>
>> You can reach the person managing the list at
>>         discuss-owner@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@hawk.iit.edu>
>> To: discuss@mpich.org
>> Subject: [mpich-discuss] beginner for remote mem access
>> Message-ID:
>>
>> <CAFNNHkwxqXUB_+b8_tZ=L5K7VcDVhsz4ZMDDMGF55v-FAhgeqg@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
>> -------------- next part --------------
>> An HTML attachment was scrubbed...
>> URL:
>> <http://lists.mpich.org/pipermail/discuss/attachments/20130703/91c4ec09/attachment-0001.html>
>>
>> ------------------------------
>>
>> Message: 2
>> Date: Wed, 3 Jul 2013 13:18:26 -0500
>> From: Yi Gu <gyi@mtu.edu>
>> To: discuss@mpich.org
>> Subject: Re: [mpich-discuss] beginner for remote mem access
>> Message-ID:
>>
>> <CAE5iO_SNJKyC9sG3XbTu4vbS=boDrOEV-Y69jUXZR6MQDanUZA@mail.gmail.com>
>> Content-Type: text/plain; charset="iso-8859-1"
>>
>> 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@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@mpich.org
>> > To manage subscription options or unsubscribe:
>> > https://lists.mpich.org/mailman/listinfo/discuss
>> >
>> -------------- next part --------------
>> An HTML attachment was scrubbed...
>> URL:
>> <http://lists.mpich.org/pipermail/discuss/attachments/20130703/4414c5cd/attachment-0001.html>
>>
>> ------------------------------
>>
>> Message: 3
>> Date: Wed, 3 Jul 2013 13:26:29 -0500
>> From: Yi Gu <gyi@mtu.edu>
>> To: discuss@mpich.org
>> Subject: Re: [mpich-discuss] beginner for remote mem access
>> Message-ID:
>>
>> <CAE5iO_Rj7uTaVsEmJ1LBHe+OpTCJjLqpMmvBmjZRd=6WLWC=EQ@mail.gmail.com>
>> Content-Type: text/plain; charset="iso-8859-1"
>>
>> 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@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@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@mpich.org
>> >> To manage subscription options or unsubscribe:
>> >> https://lists.mpich.org/mailman/listinfo/discuss
>> >>
>> >
>> >
>> -------------- next part --------------
>> An HTML attachment was scrubbed...
>> URL:
>> <http://lists.mpich.org/pipermail/discuss/attachments/20130703/362a0cda/attachment-0001.html>
>>
>> ------------------------------
>>
>> Message: 4
>> Date: Wed, 3 Jul 2013 13:28:31 -0500
>> From: Jeff Hammond <jeff.science@gmail.com>
>> To: discuss@mpich.org
>> Subject: Re: [mpich-discuss] beginner for remote mem access
>> Message-ID:
>>
>> <CAGKz=uLwki+cCY2oL_2vFwkwCi=A1Qd9VV1qXHno20gHxqY6iw@mail.gmail.com>
>> Content-Type: text/plain; charset=ISO-8859-1
>>
>> 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@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@mpich.org
>> > To manage subscription options or unsubscribe:
>> > https://lists.mpich.org/mailman/listinfo/discuss
>>
>>
>>
>> --
>> Jeff Hammond
>> jeff.science@gmail.com
>>
>>
>> ------------------------------
>>
>> _______________________________________________
>> discuss mailing list
>> discuss@mpich.org
>> https://lists.mpich.org/mailman/listinfo/discuss
>>
>> End of discuss Digest, Vol 9, Issue 5
>> *************************************
>
>
>
>
> --
> Best Regards,
> Sufeng Niu
> ECASP lab, ECE department, Illinois Institute of Technology
> Tel: 312-731-7219
>
> _______________________________________________
> discuss mailing list     discuss@mpich.org
> To manage subscription options or unsubscribe:
> https://lists.mpich.org/mailman/listinfo/discuss



--
Jeff Hammond
jeff.science@gmail.com


------------------------------

Message: 2
Date: Fri, 5 Jul 2013 12:15:22 +0100
From: anton.s.murfin@sellafieldsites.com
To: mpich-discuss@mcs.anl.gov
Subject: [mpich-discuss] Running configure on Suse Enterprise 10
Message-ID:
        <OF382230A3.E1923C9D-ON80257B9F.0038EB2D-80257B9F.003DD531@sellafieldsites.com>

Content-Type: text/plain; charset="us-ascii"

An HTML attachment was scrubbed...
URL: <http://lists.mpich.org/pipermail/discuss/attachments/20130705/d6dc7b40/attachment.html>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: c.txt
URL: <http://lists.mpich.org/pipermail/discuss/attachments/20130705/d6dc7b40/attachment.txt>

------------------------------

_______________________________________________
discuss mailing list
discuss@mpich.org
https://lists.mpich.org/mailman/listinfo/discuss

End of discuss Digest, Vol 9, Issue 7
*************************************



--
Best Regards,
Sufeng Niu
ECASP lab, ECE department, Illinois Institute of Technology
Tel: 312-731-7219