December 2014 - mpi-comments - mailman3.cels.anl.gov

(no subject)
by mustafa mohamed 19 Dec '14

19 Dec '14

Hello there ..i already do lab for mpich2 and connect all computers i have and i do cpi program and its work correctly .. but when i execute my program its give a message : the process stop working because one process exiting without calling MPI_Init .. what i can do to make it works.. ??

1 0

Re: [Mpi-comments] non-blocking communication deadlock possible?
by Maciej Szpindler 19 Dec '14

19 Dec '14

Dear Mustafa Mahamed, It sounds like your program is broken but I believe that this list is not the proper place for such issues. Try posting your question to stackoverflow.com instead. Regards, Maciej W dniu 18.12.2014 o 14:05, mustafa mohamed pisze: > Hello there ..i already do lab for mpich2 and connect all computers i have and i do cpi program and its work correctly .. but when i execute my program its give a message : the process stop working because one process exiting without calling MPI_Init .. what i can do to make it works.. ?? > > > On Wednesday, December 10, 2014 4:33 PM, Rolf Rabenseifner <rabenseifner(a)hlrs.de> wrote: > > > Dear Mr. Hehn, > > as long as you start your example with exactly two MPI processes, > your example is guaranteed to run without deadlock > based on the wording on nonblocking progress MPI-3.0 page 56, lines 30-35. > > Therefore, it is still deadlock-free if you Substitute your > >> MPI_Waitall(2,r,s); > > by > > MPI_Wait(&r[0],s[0]); > MPI_Wait(&r[1],s[1]); > > Thank you for your request on whether the MPI standard is > completely describing the nonblocking functionality. > > Best regards > Rolf > > > ----- Original Message ----- >> From: "Andreas Hehn" <hehn(a)phys.ethz.ch> >> To: mpi-comments(a)lists.mpi-forum.org >> Sent: Wednesday, December 10, 2014 12:15:28 PM >> Subject: [Mpi-comments] non-blocking communication deadlock possible? >> >> Dear MPI comments readers, >> >> I am not 100% sure if the following piece of code could cause a deadlock >> in a standard-compliant MPI implementation or not. >> >> #include <mpi.h> >> >> int main(int argc, char** argv) { >> MPI_Status s[2]; >> int num; >> >> MPI_Init(&argc, &argv); >> MPI_Comm_rank(MPI_COMM_WORLD,&num); >> >> double ds=5.4; // to send >> double dr; // to receive >> int tag=99; >> >> MPI_Request r[2]; >> if(num==0) { >> MPI_Isend(&ds,1,MPI_DOUBLE,1,tag,MPI_COMM_WORLD,&r[0]); >> MPI_Irecv (&dr,1,MPI_DOUBLE,1,tag,MPI_COMM_WORLD,&r[1]); >> } >> else { >> MPI_Isend(&ds,1,MPI_DOUBLE,0,tag,MPI_COMM_WORLD,&r[0]); >> MPI_Irecv (&dr,1,MPI_DOUBLE,0,tag,MPI_COMM_WORLD,&r[1]); >> } >> >> MPI_Waitall(2,r,s); >> >> MPI_Finalize(); >> return 0; >> } >> >> It is not entirely clear to me if the order of MPI_Isend and MPI_Irecv >> matters (as it does for blocking communication). Both function calls >> should return immediately according to section 3.7 of the standard - >> even if MPI_Isend uses synchronous send. >> So at this point there shouldn't be any problem. >> However, will MPI_Waitall() return for any order of the filed requests? >> The standard says it is equivalent to MPI_Wait() for the number of >> requests in "some arbitrary order" (line 28 of page 59). >> I suppose this means MPI_Waitall() is not supposed to mind the order of >> the requests in any way. Therefore it could not deadlock. >> Is this correct and is the line of argument sufficient to exclude the >> possibility of a deadlock? >> >> Best regards, >> >> Andreas Hehn >> _______________________________________________ >> mpi-comments mailing list >> mpi-comments(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >> > > > > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments >

1 0

Re: [Mpi-comments] non-blocking communication deadlock possible?
by mustafa mohamed 18 Dec '14

18 Dec '14

Hello there ..i already do lab for mpich2 and connect all computers i have and i do cpi program and its work correctly .. but when i execute my program its give a message : the process stop working because one process exiting without calling MPI_Init .. what i can do to make it works.. ?? On Wednesday, December 10, 2014 4:33 PM, Rolf Rabenseifner <rabenseifner(a)hlrs.de> wrote: Dear Mr. Hehn, as long as you start your example with exactly two MPI processes, your example is guaranteed to run without deadlock based on the wording on nonblocking progress MPI-3.0 page 56, lines 30-35. Therefore, it is still deadlock-free if you Substitute your > MPI_Waitall(2,r,s); by MPI_Wait(&r[0],s[0]); MPI_Wait(&r[1],s[1]); Thank you for your request on whether the MPI standard is completely describing the nonblocking functionality. Best regards Rolf ----- Original Message ----- > From: "Andreas Hehn" <hehn(a)phys.ethz.ch> > To: mpi-comments(a)lists.mpi-forum.org > Sent: Wednesday, December 10, 2014 12:15:28 PM > Subject: [Mpi-comments] non-blocking communication deadlock possible? > > Dear MPI comments readers, > > I am not 100% sure if the following piece of code could cause a deadlock > in a standard-compliant MPI implementation or not. > > #include <mpi.h> > > int main(int argc, char** argv) { > MPI_Status s[2]; > int num; > > MPI_Init(&argc, &argv); > MPI_Comm_rank(MPI_COMM_WORLD,&num); > > double ds=5.4; // to send > double dr; // to receive > int tag=99; > > MPI_Request r[2]; > if(num==0) { > MPI_Isend(&ds,1,MPI_DOUBLE,1,tag,MPI_COMM_WORLD,&r[0]); > MPI_Irecv (&dr,1,MPI_DOUBLE,1,tag,MPI_COMM_WORLD,&r[1]); > } > else { > MPI_Isend(&ds,1,MPI_DOUBLE,0,tag,MPI_COMM_WORLD,&r[0]); > MPI_Irecv (&dr,1,MPI_DOUBLE,0,tag,MPI_COMM_WORLD,&r[1]); > } > > MPI_Waitall(2,r,s); > > MPI_Finalize(); > return 0; > } > > It is not entirely clear to me if the order of MPI_Isend and MPI_Irecv > matters (as it does for blocking communication). Both function calls > should return immediately according to section 3.7 of the standard - > even if MPI_Isend uses synchronous send. > So at this point there shouldn't be any problem. > However, will MPI_Waitall() return for any order of the filed requests? > The standard says it is equivalent to MPI_Wait() for the number of > requests in "some arbitrary order" (line 28 of page 59). > I suppose this means MPI_Waitall() is not supposed to mind the order of > the requests in any way. Therefore it could not deadlock. > Is this correct and is the line of argument sufficient to exclude the > possibility of a deadlock? > > Best regards, > > Andreas Hehn > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)hlrs.de High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner Nobelstr. 19, D-70550 Stuttgart, Germany . . . . (Office: Room 1.307) _______________________________________________ mpi-comments mailing list mpi-comments(a)lists.mpi-forum.org http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

Re: [Mpi-comments] non-blocking communication deadlock possible?
by Rolf Rabenseifner 10 Dec '14

10 Dec '14

Dear Mr. Hehn, as long as you start your example with exactly two MPI processes, your example is guaranteed to run without deadlock based on the wording on nonblocking progress MPI-3.0 page 56, lines 30-35. Therefore, it is still deadlock-free if you Substitute your > MPI_Waitall(2,r,s); by MPI_Wait(&r[0],s[0]); MPI_Wait(&r[1],s[1]); Thank you for your request on whether the MPI standard is completely describing the nonblocking functionality. Best regards Rolf ----- Original Message ----- > From: "Andreas Hehn" <hehn(a)phys.ethz.ch> > To: mpi-comments(a)lists.mpi-forum.org > Sent: Wednesday, December 10, 2014 12:15:28 PM > Subject: [Mpi-comments] non-blocking communication deadlock possible? > > Dear MPI comments readers, > > I am not 100% sure if the following piece of code could cause a deadlock > in a standard-compliant MPI implementation or not. > > #include <mpi.h> > > int main(int argc, char** argv) { > MPI_Status s[2]; > int num; > > MPI_Init(&argc, &argv); > MPI_Comm_rank(MPI_COMM_WORLD,&num); > > double ds=5.4; // to send > double dr; // to receive > int tag=99; > > MPI_Request r[2]; > if(num==0) { > MPI_Isend(&ds,1,MPI_DOUBLE,1,tag,MPI_COMM_WORLD,&r[0]); > MPI_Irecv (&dr,1,MPI_DOUBLE,1,tag,MPI_COMM_WORLD,&r[1]); > } > else { > MPI_Isend(&ds,1,MPI_DOUBLE,0,tag,MPI_COMM_WORLD,&r[0]); > MPI_Irecv (&dr,1,MPI_DOUBLE,0,tag,MPI_COMM_WORLD,&r[1]); > } > > MPI_Waitall(2,r,s); > > MPI_Finalize(); > return 0; > } > > It is not entirely clear to me if the order of MPI_Isend and MPI_Irecv > matters (as it does for blocking communication). Both function calls > should return immediately according to section 3.7 of the standard - > even if MPI_Isend uses synchronous send. > So at this point there shouldn't be any problem. > However, will MPI_Waitall() return for any order of the filed requests? > The standard says it is equivalent to MPI_Wait() for the number of > requests in "some arbitrary order" (line 28 of page 59). > I suppose this means MPI_Waitall() is not supposed to mind the order of > the requests in any way. Therefore it could not deadlock. > Is this correct and is the line of argument sufficient to exclude the > possibility of a deadlock? > > Best regards, > > Andreas Hehn > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > -- Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner(a)hlrs.de High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner Nobelstr. 19, D-70550 Stuttgart, Germany . . . . (Office: Room 1.307)

1 0

non-blocking communication deadlock possible?
by Andreas Hehn 10 Dec '14

10 Dec '14

Dear MPI comments readers, I am not 100% sure if the following piece of code could cause a deadlock in a standard-compliant MPI implementation or not. #include <mpi.h> int main(int argc, char** argv) { MPI_Status s[2]; int num; MPI_Init(&argc, &argv); MPI_Comm_rank(MPI_COMM_WORLD,&num); double ds=5.4; // to send double dr; // to receive int tag=99; MPI_Request r[2]; if(num==0) { MPI_Isend(&ds,1,MPI_DOUBLE,1,tag,MPI_COMM_WORLD,&r[0]); MPI_Irecv (&dr,1,MPI_DOUBLE,1,tag,MPI_COMM_WORLD,&r[1]); } else { MPI_Isend(&ds,1,MPI_DOUBLE,0,tag,MPI_COMM_WORLD,&r[0]); MPI_Irecv (&dr,1,MPI_DOUBLE,0,tag,MPI_COMM_WORLD,&r[1]); } MPI_Waitall(2,r,s); MPI_Finalize(); return 0; } It is not entirely clear to me if the order of MPI_Isend and MPI_Irecv matters (as it does for blocking communication). Both function calls should return immediately according to section 3.7 of the standard - even if MPI_Isend uses synchronous send. So at this point there shouldn't be any problem. However, will MPI_Waitall() return for any order of the filed requests? The standard says it is equivalent to MPI_Wait() for the number of requests in "some arbitrary order" (line 28 of page 59). I suppose this means MPI_Waitall() is not supposed to mind the order of the requests in any way. Therefore it could not deadlock. Is this correct and is the line of argument sufficient to exclude the possibility of a deadlock? Best regards, Andreas Hehn

1 0

Re: [Mpi-comments] Error in MPI_Neighbor_alltoallw example?
by Timothy Brown 05 Dec '14

05 Dec '14

Ah, perfect thanks! I didn't know about the trac/ticket system. Tim > On Dec 5, 2014, at 9:04 AM, Rajeev Thakur <thakur(a)mcs.anl.gov> wrote: > > Yes, see https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/420 > > Rajeev > > > On Dec 5, 2014, at 9:41 AM, Timothy Brown <Timothy.Brown-1(a)Colorado.EDU> wrote: > >> Hello. >> >> I was trying to use MPI_Neighbor_alltoallw() and was following the example on page 333 of the MPI-3.0 standard. However I believe the send and receive counts are incorrect. Lines 21 to 25: >> >> DO i=1,num_neigh >> sndcounts(i) = 100 >> rcvcounts(i) = 100 >> rcvtypes(i) = sndtypes(i) >> END DO >> >> Since the send/receive counts for 1 and 2 are vectors types of length 100, we should be sending/receiving only 1 type/element. As follows: >> >> sndcounts(1) = 1 >> sndcounts(2) = 1 >> sndcounts(3) = 100 >> sndcounts(4) = 100 >> >> rcvcounts = sndcounts >> >> Indeed if I compile the example as per the standard, I get data corruption. >> >> I couldn't find this in the errata published on May 14, 2014. However if it is a known mistake please ignore my comment. >> >> Regards >> Timothy Brown >> _______________________________________________ >> mpi-comments mailing list >> mpi-comments(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments > > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

Re: [Mpi-comments] Error in MPI_Neighbor_alltoallw example?
by Rajeev Thakur 05 Dec '14

05 Dec '14

Yes, see https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/420 Rajeev On Dec 5, 2014, at 9:41 AM, Timothy Brown <Timothy.Brown-1(a)Colorado.EDU> wrote: > Hello. > > I was trying to use MPI_Neighbor_alltoallw() and was following the example on page 333 of the MPI-3.0 standard. However I believe the send and receive counts are incorrect. Lines 21 to 25: > > DO i=1,num_neigh > sndcounts(i) = 100 > rcvcounts(i) = 100 > rcvtypes(i) = sndtypes(i) > END DO > > Since the send/receive counts for 1 and 2 are vectors types of length 100, we should be sending/receiving only 1 type/element. As follows: > > sndcounts(1) = 1 > sndcounts(2) = 1 > sndcounts(3) = 100 > sndcounts(4) = 100 > > rcvcounts = sndcounts > > Indeed if I compile the example as per the standard, I get data corruption. > > I couldn't find this in the errata published on May 14, 2014. However if it is a known mistake please ignore my comment. > > Regards > Timothy Brown > _______________________________________________ > mpi-comments mailing list > mpi-comments(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-comments

1 0

Error in MPI_Neighbor_alltoallw example?
by Timothy Brown 05 Dec '14

05 Dec '14

Hello. I was trying to use MPI_Neighbor_alltoallw() and was following the example on page 333 of the MPI-3.0 standard. However I believe the send and receive counts are incorrect. Lines 21 to 25: DO i=1,num_neigh sndcounts(i) = 100 rcvcounts(i) = 100 rcvtypes(i) = sndtypes(i) END DO Since the send/receive counts for 1 and 2 are vectors types of length 100, we should be sending/receiving only 1 type/element. As follows: sndcounts(1) = 1 sndcounts(2) = 1 sndcounts(3) = 100 sndcounts(4) = 100 rcvcounts = sndcounts Indeed if I compile the example as per the standard, I get data corruption. I couldn't find this in the errata published on May 14, 2014. However if it is a known mistake please ignore my comment. Regards Timothy Brown

1 0