[mpich-discuss] Memory leaks using ROMIO
Akram Smaali
akram.smaali at gmail.com
Fri Jul 13 02:55:15 CDT 2012
---------- Forwarded message ----------
From: Akram Smaali <akram.smaali at gmail.com>
Date: 2012/7/12
Subject: Re: [mpich-discuss] Memory leaks using ROMIO
To: Jayesh Krishna <jayesh at mcs.anl.gov>
I think that, as Rob latham said, it's caused by ntfs. I was looking into
the source code of ROMIO but i didn't see anything indicating that. I
suppose that the OS doesn't get back the allocated buffer from the user
space.
# Does the system memory decrease after a period of time?
No, it does just at the end.
# Does the memory increase correspond to the amount of data written?
Yes excactly the amount of data written.
# How much does the memory increase?
i use a 5GB file so it increase that much.
# Do you see similar behavior with a simple Windows program
(CreateFileMapping() + WriteFile() + CloseHandle())?
It just happens with the MPI_File_read_at()
2012/7/12 Jayesh Krishna <jayesh at mcs.anl.gov>
> Hi,
> Windows is known to delay writes even after the user is done with the
> file (I have seen similar issues before. Your problem could be different
> though.).
>
> # Does the system memory decrease after a period of time?
> # Does the memory increase correspond to the amount of data written?
> # How much does the memory increase?
> # Do you see similar behavior with a simple Windows program
> (CreateFileMapping() + WriteFile() + CloseHandle())?
>
> Regards,
> Jayesh
>
> ----- Original Message -----
> From: "Rob Latham" <robl at mcs.anl.gov>
> To: mpich-discuss at mcs.anl.gov
> Sent: Thursday, July 12, 2012 11:09:50 AM
> Subject: Re: [mpich-discuss] Memory leaks using ROMIO
>
> On Thu, Jul 12, 2012 at 01:29:38PM +0200, Akram Smaali wrote:
> > Hello,
> >
> > i'm using mpich2 in a C program that reads several buffers from a data
> text
> > files.
> > I noticed that even when i free the buffer at the end, the global memory
> > usage still increase but the memory usage per process still the same.
> > I changed the MPI_File_read_at() with fseek/fread and the problem has
> > dissapeared.
> > I'm wondering if that i should do some extra instriuctions to avoid that
> > because it seems that mpi_read allocate an intermediate buffer for
> reading
> > (sieving optimization maybe) from the user space memory and the OS (
> > windosws 7) don't get it back.
> > i'm using MPICH2 1.4 on w windows 7 with NTFS file system.
>
> It might be an NTFS-specific leak. I don't see any leaks on my linux
> system (ad_ufs).
>
> > Thank you for your help.
> >
> > here is my code :
>
> It would be great if you could make this case more self contained.
> Whatever offsets are in your Dictionary file, put those in an array
> and broadcast that.
>
> > #include <stdio.h>
> > #include <stdlib.h>
> > #include <mpi.h>
> > #include <omp.h>
> > #include <string.h>
> > #include <time.h>
> > #include "queue.h"
> > #include "KhiopsNativeInterface.h"
> > #include <windows.h>
> >
> > #define maxLength 100000
> >
> > double CPUtime(){ return ((double) clock())/CLOCKS_PER_SEC;}
> >
> >
> > int main(int argc, char* argv []){
> >
> > if(argc != 5) {
> > printf("\t[Dictionary file] [Dictionary] [Input file] [Buffer size]\n");
> > exit(0);
> > }
> >
> >
> >
> > char* sDictionaryFileName = argv[1];
> > char* sDictionaryName = argv[2];
> > char* filename = argv[3];
> > int Mbuffer = atoi(argv[4]);
> >
> > int maxBuffer = Mbuffer*1024*1024;
> > int over = 10000;
> >
> > int rank,numprocess;
> > long int offset;
> >
> > char* buffer;
> > char* opbuffer;
> >
> > double tstart=CPUtime();
> >
> > MPI_Init( &argc, &argv );
> > MPI_Comm_rank( MPI_COMM_WORLD, &rank );
> >
> > /* mpi version */
> > /* open the file*/
> > MPI_File fh;
> > int err;
> > err = MPI_File_open(MPI_COMM_WORLD, filename, MPI_MODE_RDONLY,
> > MPI_INFO_NULL, &fh);
> > if (err != MPI_SUCCESS) {
> > char errstr[MPI_MAX_ERROR_STRING];
> > int errlen;
> > MPI_Error_string(err, errstr, &errlen);
> > printf("Error at opening file %s (%s)\n",filename,errstr);
> > MPI_Finalize();
> > exit(1);
> > }
> > // get offsets and buffer size
> > MPI_Offset sfile;
> > MPI_File_get_size(fh,&sfile);
> > MPI_Status status;
> >
> >
> > /* C version */
> > /*FILE* fh;
> > long int sfile;
> > fh =fopen( filename,"rb");
> > if (fh==NULL) {
> > printf("Error at opening file %s\n",filename);
> > exit(1);
> > }
> > // get offsets and buffer size
> > fseek(fh, 0L, SEEK_END);
> > sfile = ftell(fh);
> > fseek(fh, 0L, SEEK_SET);*/
> >
> >
> > MPI_Comm_size( MPI_COMM_WORLD, &numprocess );
> >
> > /* number of iterations */
> > long int data_size = (long int)(sfile/(numprocess));
> > int nbIter = data_size/maxBuffer;
> > if(nbIter<=1){
> > nbIter = 1;
> > maxBuffer = data_size;
> > }
> >
> > /* offsets */
> > offset = data_size*(rank);
> > long int cursor = offset;
> > char* header;
> > if(rank==0){
> > FILE* fh;
> > fh =fopen( filename,"rb");
> > if (fh==NULL) {
> > printf("Error at opening file %s\n",filename);
> > exit(1);
> > }
> > /* read the header and broadcast it */
> > header = malloc(sizeof(char)*1000);
> > fgets(header,1000,fh);
> > fclose(fh);
> >
> > //broadcast header
> > int sndHeader = strlen(header);
> > //cursor+=sndHeader;
> > int process_counter;
> > for(process_counter=1;process_counter<numprocess;process_counter++){
> > int ierr = MPI_Send(&sndHeader,sizeof(int), MPI_INT, process_counter,
> > 42,MPI_COMM_WORLD);
> > if (ierr != MPI_SUCCESS) {
> > int errclass,resultlen;
> > char err_buffer[MPI_MAX_ERROR_STRING];
> > MPI_Error_class(ierr,&errclass);
> > if (errclass== MPI_ERR_RANK) {
> > fprintf(stderr,"Invalid rank used in MPI send call\n");
> > MPI_Error_string(ierr,err_buffer,&resultlen);
> > fprintf(stderr,err_buffer);
> > MPI_Finalize();
> > }
> > }
> > MPI_Send(header, sndHeader, MPI_CHAR, process_counter, 43,
> MPI_COMM_WORLD);
> > }
> > }
> > else{
> > /* receive the header */
> > int sizeofHeader;
> > MPI_Status s ;
> > MPI_Recv(&sizeofHeader,sizeof(int),MPI_INT,0,42,MPI_COMM_WORLD,&s);
> > header = malloc (sizeof(char)*sizeofHeader+1);
> > MPI_Recv(header,sizeofHeader,MPI_CHAR,0,43,MPI_COMM_WORLD,&s);
> > }
> >
> >
> >
> > /* Synchronization barrier */
> > MPI_Barrier(MPI_COMM_WORLD);
> >
> > int count;
> >
> > opbuffer = malloc(sizeof(char)*maxBuffer);
> >
> > /* C version */
> > //fseek(fh,cursor,SEEK_SET);
> >
> > for(count=0;count<nbIter;count++){
> >
> > if(count==0 && rank==numprocess-1){ //init ring
> > //send the token to p0
> > int token=1;
> > MPI_Send(&token,sizeof(int),MPI_INT,0,55,MPI_COMM_WORLD);
> > }
> >
> > //recv
> > int token;
> > int sender;
> > if(rank==0)
> > sender = numprocess-1;
> > else
> > sender=rank-1;
> >
> > MPI_Status s;
> > MPI_Recv(&token,sizeof(int),MPI_INT,sender,55,MPI_COMM_WORLD,&s);
> > fflush(stdout);printf("P%d got the token at %G\n",rank,CPUtime());
> > //read
> > double start=CPUtime();
> > /*double readtime;
> > double sread=CPUtime();//read time*/
> >
> > //read
> > if(token==1){
> > /* MPI version */
> > int err=MPI_File_read_at(fh, cursor,opbuffer, sizeof(char)*maxBuffer,
> > MPI_CHAR, &status);
> > if(err!=MPI_SUCCESS){
> > char errstr[MPI_MAX_ERROR_STRING];
> > int errlen;
> > MPI_Error_string(err, errstr, &errlen);
> > printf("Error reading file %s (%s)\n",filename,errstr);
> > MPI_Finalize();
> > exit(0);
> > }
> >
> > /* C version of read */
> > /*int k=fread(opbuffer,sizeof(char),maxBuffer,fh);
> > if(k==0)
> > perror("fread");*/
> >
> > cursor+=maxBuffer;
> > buffer=opbuffer;
> >
> > }
> > else{
> > printf("Error token!\n");
> > token=1;
> > }
> > //printf("P%d readtime=%G\n",rank,CPUtime()-sread);
> > //Isend
> > int next = (rank+1)%numprocess;
> > MPI_Send(&token,sizeof(int),MPI_INT,next,55,MPI_COMM_WORLD);
> >
> > //i perform processing buffer here!
> >
> > }
> > free(opbuffer);
> >
> > /* mpi version */
> > int er=MPI_File_close(&fh);
> > if(er!=MPI_SUCCESS){
> > printf("Error closing file\n");
> > MPI_Finalize();
> > exit(1);
> > }
> >
> > /* c version */
> > //fclose(fh);
> >
> >
> > MPI_Finalize();
> >
> > printf("Global time : %G\n",CPUtime()-tstart);
> > return 0;
> > }
> >
> >
> > Best Regards
>
> > _______________________________________________
> > mpich-discuss mailing list mpich-discuss at mcs.anl.gov
> > To manage subscription options or unsubscribe:
> > https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>
>
> --
> Rob Latham
> Mathematics and Computer Science Division
> Argonne National Lab, IL USA
> _______________________________________________
> mpich-discuss mailing list mpich-discuss at mcs.anl.gov
> To manage subscription options or unsubscribe:
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>
--
*Akram SMAALI*
Ingénieur stagiaire à France Télécom R&D
22300 Lannion, France
*mobile: +33 7 51 13 39 11 *
*mail:* akram.smaali at gmail.com
--
*Akram SMAALI*
Ingénieur stagiaire à France Télécom R&D
22300 Lannion, France
*mobile: +33 7 51 13 39 11 *
*mail:* akram.smaali at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20120713/02e2fb70/attachment-0001.html>
More information about the mpich-discuss
mailing list