users-boun...@open-mpi.org a écrit sur 19/04/2012 10:24:16 :
> De : Rohan Deshpande <rohan...@gmail.com>
> A : Open MPI Users <us...@open-mpi.org>
> Date : 19/04/2012 10:24
> Objet : Re: [OMPI users] machine exited on signal 11 (Segmentation
fault).
> Envoyé par : users-boun...@open-mpi.org
>
> Hi Pascal,
>
> The offset is received from the master task. so no need to
> initialize for non master tasks?
>
> not sure what u meant exactly.
>
This is OK.
I have two remarks:
1) In the update() subroutine, you declare a table of int of size
"myoffset+chunk" that will be allocated in the process stack.
For the task number 2, the size of this table is 4*3*1,000,000 = 12
Mbytes
For the task number 3, the size of this table is 4*4*1,000,000 = 16
Mbytes
This huge table is not used. Try to delete the declaration (or comment
it).
2) When you increase the size of ARRAYSIZE = 20,000,000, the value of sum
in the first process should be:
(10**6)*(10**6+1)/2. That's approximatly 10**12 or 2**40. This cannot
be calcultated with the variable sum (int).
Hope this will help.
> Thanks
>
>
> On Thu, Apr 19, 2012 at 3:36 PM, <pascal.dev...@bull.net> wrote:
> I do not see where you initialize the offset on the "Non-master tasks
> ". This could be the problem.
>
> Pascal
>
> users-boun...@open-mpi.org a écrit sur 19/04/2012 09:18:31 :
>
> > De : Rohan Deshpande <rohan...@gmail.com>
> > A : Open MPI Users <us...@open-mpi.org>
> > Date : 19/04/2012 09:18
> > Objet : Re: [OMPI users] machine exited on signal 11 (Segmentation
fault).
> > Envoyé par : users-boun...@open-mpi.org
> >
> > Hi Jeffy,
> >
> > I checked the SEND RECV buffers and it looks ok to me. The code I
> > have sent works only when I statically initialize the array.
> >
> > The code fails everytime I use malloc to initialize the array.
> >
> > Can you please look at code and let me know what is wrong.
>
> > On Wed, Apr 18, 2012 at 8:11 PM, Jeffrey Squyres <jsquy...@cisco.com>
wrote:
> > As a guess, you're passing in a bad address.
> >
> > Double check the buffers that you're sending to MPI_SEND/MPI_RECV/etc.
> >
> >
> > On Apr 17, 2012, at 10:43 PM, Rohan Deshpande wrote:
> >
> > > After using malloc i am getting following error
> > >
> > > *** Process received signal ***
> > > Signal: Segmentation fault (11)
> > > Signal code: Address not mapped (1)
> > > Failing at address: 0x1312d08
> > > [ 0] [0x5e840c]
> > > [ 1] /usr/local/lib/openmpi/mca_btl_tcp.so(+0x5bdb) [0x119bdb]
> > > /usr/local/lib/libopen-pal.so.0(+0x19ce0) [0xb2cce0]
> > > /usr/local/lib/libopen-pal.so.0(opal_event_loop+0x27) [0xb2cf47]
> > > /usr/local/lib/libopen-pal.so.0(opal_progress+0xda) [0xb200ba]
> > > /usr/local/lib/openmpi/mca_pml_ob1.so(+0x3f75) [0xa9ef75]
> > > [ 6] /usr/local/lib/libmpi.so.0(MPI_Recv+0x136) [0xea7c46]
> > > [ 7] mpi_array(main+0x501) [0x8048e25]
> > > [ 8] /lib/libc.so.6(__libc_start_main+0xe6) [0x2fece6]
> > > [ 9] mpi_array() [0x8048891]
> > > *** End of error message ***
> > > [machine4][[3968,1],0][btl_tcp_frag.c:216:mca_btl_tcp_frag_recv]
> > mca_btl_tcp_frag_recv: readv failed: Connection reset by peer (104)
> > >
--------------------------------------------------------------------------
> > > mpirun noticed that process rank 1 with PID 2936 on node machine4
> > exited on signal 11 (Segmentation fault).
> > >
> > > Can someone help please.
> > >
> > > Thanks
> > >
> > >
> > >
> > > On Tue, Apr 17, 2012 at 6:01 PM, Jeffrey Squyres <jsquy...@cisco.com
> > wrote:
> > > Try malloc'ing your array instead of creating it statically on the
> > stack. Something like:
> > >
> > > int *data;
> > >
> > > int main(..) {
> > > {
> > > data = malloc(ARRAYSIZE * sizeof(int));
> > > if (NULL == data) {
> > > perror("malloc");
> > > exit(1);
> > > }
> > > // ...
> > > }
> > >
> > >
> > > On Apr 17, 2012, at 5:05 AM, Rohan Deshpande wrote:
> > >
> > > >
> > > > Hi,
> > > >
> > > > I am trying to distribute large amount of data using MPI.
> > > >
> > > > When I exceed the certain data size the segmentation fault occurs.
> > > >
> > > > Here is my code,
> > > >
> > > >
> > > > #include "mpi.h"
> > > > #include <stdio.h>
> > > > #include <stdlib.h>
> > > > #include <string.h>
> > > > #define ARRAYSIZE 2000000
> > > > #define MASTER 0
> > > >
> > > > int data[ARRAYSIZE];
> > > >
> > > >
> > > > int main(int argc, char* argv[])
> > > > {
> > > > int numtasks, taskid, rc, dest, offset, i, j, tag1, tag2,
> > source, chunksize, namelen;
> > > > int mysum, sum;
> > > > int update(int myoffset, int chunk, int myid);
> > > > char myname[MPI_MAX_PROCESSOR_NAME];
> > > > MPI_Status status;
> > > > double start, stop, time;
> > > > double totaltime;
> > > > FILE *fp;
> > > > char line[128];
> > > > char element;
> > > > int n;
> > > > int k=0;
> > > >
> > > >
> > > >
> > > > /***** Initializations *****/
> > > > MPI_Init(&argc, &argv);
> > > > MPI_Comm_size(MPI_COMM_WORLD, &numtasks);
> > > > MPI_Comm_rank(MPI_COMM_WORLD,&taskid);
> > > > MPI_Get_processor_name(myname, &namelen);
> > > > printf ("MPI task %d has started on host %s...\n", taskid,
myname);
> > > > chunksize = (ARRAYSIZE / numtasks);
> > > > tag2 = 1;
> > > > tag1 = 2;
> > > >
> > > >
> > > > /***** Master task only ******/
> > > > if (taskid == MASTER){
> > > >
> > > > /* Initialize the array */
> > > > sum = 0;
> > > > for(i=0; i<ARRAYSIZE; i++) {
> > > > data[i] = i * 1 ;
> > > > sum = sum + data[i];
> > > > }
> > > > printf("Initialized array sum = %d\n",sum);
> > > >
> > > > /* Send each task its portion of the array - master keeps 1st
part */
> > > > offset = chunksize;
> > > > for (dest=1; dest<numtasks; dest++) {
> > > > MPI_Send(&offset, 1, MPI_INT, dest, tag1, MPI_COMM_WORLD);
> > > > MPI_Send(&data[offset], chunksize, MPI_INT, dest, tag2,
> > MPI_COMM_WORLD);
> > > > printf("Sent %d elements to task %d offset= %d
> > \n",chunksize,dest,offset);
> > > > offset = offset + chunksize;
> > > > }
> > > >
> > > > /* Master does its part of the work */
> > > > offset = 0;
> > > > mysum = update(offset, chunksize, taskid);
> > > >
> > > > /* Wait to receive results from each task */
> > > > for (i=1; i<numtasks; i++) {
> > > > source = i;
> > > > MPI_Recv(&offset, 1, MPI_INT, source, tag1,
> MPI_COMM_WORLD, &status);
> > > > MPI_Recv(&data[offset], chunksize, MPI_INT, source, tag2,
> > > > MPI_COMM_WORLD, &status);
> > > > }
> > > >
> > > > /* Get final sum and print sample results */
> > > > MPI_Reduce(&mysum, &sum, 1, MPI_INT, MPI_SUM, MASTER,
MPI_COMM_WORLD);
> > > > /* printf("Sample results: \n");
> > > > offset = 0;
> > > > for (i=0; i<numtasks; i++) {
> > > > for (j=0; j<5; j++)
> > > > printf(" %d",data[offset+j]);ARRAYSIZE
> > > > printf("\n");
> > > > offset = offset + chunksize;
> > > > }*/
> > > > printf("\n*** Final sum= %d ***\n",sum);
> > > >
> > > > } /* end of master section */
> > > >
> > > >
> > > > #include <stdlib.h>
> > > > /***** Non-master tasks only *****/
> > > >
> > > > if (taskid > MASTER) {
> > > >
> > > > /* Receive my portion of array from the master task */
> > > > start= MPI_Wtime();
> > > > source = MASTER;
> > > > MPI_Recv(&offset, 1, MPI_INT, source, tag1, MPI_COMM_WORLD,
&status);
> > > > MPI_Recv(&data[offset], chunksize, MPI_INT, source,
> > tag2,MPI_COMM_WORLD, &status);
> > > >
> > > > mysum = update(offset, chunksize, taskid);
> > > > stop = MPI_Wtime();
> > > > time = stop -start;
> > > > printf("time taken by process %d to recieve elements and
> > caluclate own sum is = %lf seconds \n", taskid, time);
> > > > totaltime = totaltime + time;
> > > >
> > > > /* Send my results back to the master task */
> > > > dest = MASTER;
> > > > MPI_Send(&offset, 1, MPI_INT, dest, tag1, MPI_COMM_WORLD);
> > > > MPI_Send(&data[offset], chunksize, MPI_INT, MASTER, tag2,
> > MPI_COMM_WORLD);
> > > >
> > > > MPI_Reduce(&mysum, &sum, 1, MPI_INT, MPI_SUM, MASTER,
MPI_COMM_WORLD);
> > > >
> > > > } /* end of non-master */
> > > >
> > > > // printf("Total time taken for distribution is - %lf
> > seconds", totaltime);
> > > > MPI_Finalize();
> > > >
> > > > } /* end of main */
> > > >
> > > >
> > > > int update(int myoffset, int chunk, int myid) {
> > > > int i,j;
> > > > int mysum;
> > > > int mydata[myoffset+chunk];
> > > > /* Perform addition to each of my array elements and keep my sum
*/
> > > > mysum = 0;
> > > > /* printf("task %d has elements:",myid);
> > > > for(j = myoffset; j<myoffset+chunk; j++){
> > > > printf("\t%d", data[j]);
> > > > }
> > > > printf("\n");*/
> > > > for(i=myoffset; i < myoffset + chunk; i++) {
> > > >
> > > > //data[i] = data[i] + i;
> > > > mysum = mysum + data[i];
> > > > }
> > > > printf("Task %d has sum = %d\n",myid,mysum);
> > > > return(mysum);
> > > > }
> > > >
> > > >
> > > > When I run it with ARRAYSIZE = 2000000 The program works fine.
> > But when I increase the size ARRAYSIZE = 20000000. The program ends
> > with segmentation fault.
> > > > I am running it on a cluster (machine 4 is master, machine 5,6
> > are slaves) and np=20
> > > >
> > > > MPI task 0 has started on host machine4
> > > > MPI task 2 has started on host machine4
> > > > MPI task 3 has started on host machine4
> > > > MPI task 14 has started on host machine4
> > > > MPI task 8 has started on host machine6
> > > > MPI task 10 has started on host machine6
> > > > MPI task 13 has started on host machine4
> > > > MPI task 4 has started on host machine5
> > > > MPI task 6 has started on host machine5
> > > > MPI task 7 has started on host machine5
> > > > MPI task 16 has started on host machine5
> > > > MPI task 11 has started on host machine6
> > > > MPI task 12 has started on host machine4
> > > > MPI task 5 has started on hostmachine5
> > > > MPI task 17 has started on host machine5
> > > > MPI task 18 has started on host machine5
> > > > MPI task 15 has started on host machine4
> > > > MPI task 19 has started on host machine5
> > > > MPI task 1 has started on host machine4
> > > > MPI task 9 has started on host machine6
> > > > Initialized array sum = 542894464
> > > > Sent 1000000 elements to task 1 offset= 1000000
> > > > Task 1 has sum = 1055913696
> > > > time taken by process 1 to recieve elements and caluclate own
> > sum is = 0.249345 seconds
> > > > Sent 1000000 elements to task 2 offset= 2000000
> > > > Sent 1000000 elements to task 3 offset= 3000000
> > > > Task 2 has sum = 328533728
> > > > time taken by process 2 to recieve elements and caluclate own
> > sum is = 0.274285 seconds
> > > > Sent 1000000 elements to task 4 offset= 4000000
> > > >
>
--------------------------------------------------------------------------
> > > > mpirun noticed that process rank 3 with PID 5695 on node
> > machine4 exited on signal 11 (Segmentation fault).
> > > >
> > > > Any idea what could be wrong here?
> > > >
> > > >
> > > > --
> > > >
> > > > Best Regards,
> > > >
> > > > ROHAN DESHPANDE
> > > >
> > > >
> > > >
> > > > _______________________________________________
> > > > users mailing list
> > > > us...@open-mpi.org
> > > > http://www.open-mpi.org/mailman/listinfo.cgi/users
> > >
> > >
> > > --
> > > Jeff Squyres
> > > jsquy...@cisco.com
> > > For corporate legal information go to: http://www.cisco.com/web/
> > about/doing_business/legal/cri/
> > >
> > >
> > > _______________________________________________
> > > users mailing list
> > > us...@open-mpi.org
> > > http://www.open-mpi.org/mailman/listinfo.cgi/users
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > _______________________________________________
> > > users mailing list
> > > us...@open-mpi.org
> > > http://www.open-mpi.org/mailman/listinfo.cgi/users
> >
> >
> > --
> > Jeff Squyres
> > jsquy...@cisco.com
> > For corporate legal information go to: http://www.cisco.com/web/
> > about/doing_business/legal/cri/
> >
> >
> > _______________________________________________
> > users mailing list
> > us...@open-mpi.org
> > http://www.open-mpi.org/mailman/listinfo.cgi/users
> >
> >
> >
> > --
> >
> > Best Regards,
> >
> > ROHAN DESHPANDE
> >
>
> > _______________________________________________
> > users mailing list
> > us...@open-mpi.org
> > http://www.open-mpi.org/mailman/listinfo.cgi/users
>
> _______________________________________________
> users mailing list
> us...@open-mpi.org
> http://www.open-mpi.org/mailman/listinfo.cgi/users
>
>
>
> --
>
> Best Regards,
>
> ROHAN DESHPANDE
>
> _______________________________________________
> users mailing list
> us...@open-mpi.org
> http://www.open-mpi.org/mailman/listinfo.cgi/users
