MPI in C

ltee · 03-26-2009, 05:15 AM

Hi all

I am using MPI and c programming. i want to collect strings from processors to root processor
now i have writen the code as below but does not display the strings. Can anyone help me with this?
The code is as below:

Code:

#include <mpi.h>
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <string.h>
int main(int argc, char *argv[]){
int numProcs;
int myRank, a, *data, i;
char **DATA;
char *b;
//initialize MPI
MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD,&myRank);
MPI_Comm_size(MPI_COMM_WORLD, &numProcs);
// do some computation and MPI communications
a = 1000+myRank;
 b = "sdfsd";
if (myRank == 1){
data = (int *) malloc(numProcs*sizeof(int));
DATA = (char **) calloc(numProcs, sizeof(char **));
for (i=0;i<numProcs;i++){
data[i] = 0;
DATA[i] = "helloP";
 }
printf("\n contents of data array before gather:\n");
for (i=0;i<numProcs;i++)
printf("%s\n", DATA[i]);

}

MPI_Gather(&b, 5, MPI_CHAR, DATA, 5,  MPI_CHAR,1, MPI_COMM_WORLD);

printf("here are contents %s\n",DATA[1]); 
if (myRank == 1){
printf("\n contents of data array after gather:\n");
for (i=0;i<numProcs;i++){
	printf("%s\n", DATA[i]);
 /*printf("2. DATA[%d] = %s \n",i, DATA[i]);*/}}

//shutdown MPI
MPI_Finalize();
return 0;
}

matsp · 03-26-2009, 05:18 AM

Don't see anything DIRECTLY wrong. What are you actually receiving?

--
Mats

ltee · 03-26-2009, 05:31 AM

Thanks matsp for quick response. Actually bellow is the message i receive:

rank 2 in job 8 cslab117-002_50788 caused collective abort of all ranks
exit status of rank 2: killed by signal 11
rank 0 in job 8 cslab117-002_50788 caused collective abort of all ranks
exit status of rank 0: killed by signal 11

matsp · 03-26-2009, 05:49 AM

signal 11 is a "seg-fault", which means your app tried to write to memory that doesn't exist.

I suspect it is because your DATA is actually pointing to read-only memory. Maybe you would be better off allocating some memory for DATA[i] instead of setting it to "HELLO". I don't KNOW that this is the problem, but it seems reasonable that your processes can not write to read-only memory, and that would cause a Seg-fault signal.

--
Mats

ltee · 03-26-2009, 06:05 AM

well, when i remove the line

Code:

 DATA[i] = "helloP";

i still get the message:
rank 1 in job 11 cslab117-002_50788 caused collective abort of all ranks
exit status of rank 1: killed by signal 11
Clear allocation of memory is not a problem here. so im wondering exactly what i need to change. If you look at line:

Code:

 MPI_Gather(&b, 5, MPI_CHAR, DATA, 5,  MPI_CHAR,1, MPI_COMM_WORLD);

, can you pick something that might be causing this error, because i suspect that the error is there

matsp · 03-26-2009, 06:10 AM

So where is DATA[i] pointing now? Just removing the assignment IS NOT ENOUGH. You need each process to have some space to write the data into.

--
Mats

03-26-2009, 05:18 AM	#2
matsp Kernel hacker Join Date: Jul 2007 Location: Farncombe, Surrey, England Posts: 15,686	Don't see anything DIRECTLY wrong. What are you actually receiving? -- Mats __________________ Compilers can produce warnings - make the compiler programmers happy: Use them! Please don't PM me for help - and no, I don't do help over instant messengers.
matsp is offline

03-26-2009, 05:31 AM	#3
ltee Registered User Join Date: Mar 2009 Posts: 3	Thanks matsp for quick response. Actually bellow is the message i receive: rank 2 in job 8 cslab117-002_50788 caused collective abort of all ranks exit status of rank 2: killed by signal 11 rank 0 in job 8 cslab117-002_50788 caused collective abort of all ranks exit status of rank 0: killed by signal 11
ltee is offline

03-26-2009, 05:49 AM	#4
matsp Kernel hacker Join Date: Jul 2007 Location: Farncombe, Surrey, England Posts: 15,686	signal 11 is a "seg-fault", which means your app tried to write to memory that doesn't exist. I suspect it is because your DATA is actually pointing to read-only memory. Maybe you would be better off allocating some memory for DATA[i] instead of setting it to "HELLO". I don't KNOW that this is the problem, but it seems reasonable that your processes can not write to read-only memory, and that would cause a Seg-fault signal. -- Mats __________________ Compilers can produce warnings - make the compiler programmers happy: Use them! Please don't PM me for help - and no, I don't do help over instant messengers.
matsp is offline

03-26-2009, 06:05 AM	#5
ltee Registered User Join Date: Mar 2009 Posts: 3	well, when i remove the line Code: DATA[i] = "helloP"; i still get the message: rank 1 in job 11 cslab117-002_50788 caused collective abort of all ranks exit status of rank 1: killed by signal 11 Clear allocation of memory is not a problem here. so im wondering exactly what i need to change. If you look at line: Code: MPI_Gather(&b, 5, MPI_CHAR, DATA, 5, MPI_CHAR,1, MPI_COMM_WORLD); , can you pick something that might be causing this error, because i suspect that the error is there
ltee is offline

03-26-2009, 06:10 AM	#6
matsp Kernel hacker Join Date: Jul 2007 Location: Farncombe, Surrey, England Posts: 15,686	So where is DATA[i] pointing now? Just removing the assignment IS NOT ENOUGH. You need each process to have some space to write the data into. -- Mats __________________ Compilers can produce warnings - make the compiler programmers happy: Use them! Please don't PM me for help - and no, I don't do help over instant messengers.
matsp is offline

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Communication using MPI	Cell	Linux Programming	9	08-13-2009 02:28 AM
MPI, which way will be best?	maverick_starst	C++ Programming	5	06-30-2009 02:17 PM
Sorting whit MPI	isato	C Programming	0	03-03-2009 10:38 AM
Malloc and MPI	moddinati	C Programming	17	03-07-2008 07:55 PM
MPI programming	kris.c	Tech Board	1	12-08-2006 12:25 PM