Re: [OMPI users] Put/Get semantics
Instead of MPI_Alloc_mem and MPI_Win_create, you should use MPI_Win_allocate. This will make it much easier for the implementation to optimize with interprocess shared memory and exploit scalability features such as symmetric globally addressable memory. It also obviates the need to do both MPI_Win_free and MPI_Free_mem. Based upon what I've seen recently ( https://travis-ci.org/jeffhammond/armci-mpi), using MPI_Win_allocate may fix some unresolved Open-MPI RMA bugs ( https://github.com/open-mpi/ompi/issues/1275). As for your synchronization question, instead of MPI_Rget(b,1,dtype,rproc,displ,1,dtype,win,&request); MPI_Wait(&request,&status); and MPI_Rput(a,1,dtype,rproc,displ,1,dtype,win,&request); MPI_Wait(&request,&status); you should use MPI_Get(b,1,dtype,rproc,displ,1,dtype,win); MPI_Win_flush_local(1,win); and MPI_Put(a,1,dtype,rproc,displ,1,dtype,win); MPI_Win_flush_local(1,win); as there is no need to create a request for this usage model. Request-based RMA entails some implementation overhead in some cases, and is more likely to be broken since it is not heavily tested. On the other hand, the non-request RMA has been tested extensively thanks to the thousands of NWChem jobs I've run using ARMCI-MPI on Cray, InfiniBand, and other systems. As I think I've said before on some list, one of the best ways to understand the mapping between ARMCI and MPI RMA is to look at ARMCI-MPI. Jeff On Wed, Jan 6, 2016 at 8:51 AM, Palmer, Bruce J wrote: > > Hi, > I’m trying to compare the semantics of MPI RMA with those of ARMCI. I’ve written a small test program that writes data to a remote processor and then reads the data back to the original processor. In ARMCI, you should be able to do this since operations to the same remote processor are completed in the same order that they are requested on the calling processor. I’ve implemented this two different ways using MPI RMA. The first is to call MPI_Win_lock to create a shared lock on the remote processor, then MPI_Put/MPI_Get to initiate the data transfer and finally MPI_Win_unlock to force completion of the data transfer. My understanding is that this should allow you to write and then read data to the same process, since the first triplet > MPI_Win_lock > MPI_Put > MPI_Win_unlock > must be completed both locally and remotely before the unlock call completes. The calls in the second triplet > MPI_Win_lock > MPI_Get > MPI_Win_unlock > cannot start until the first triplet is done, so if both the put and the get refer to the same data on the same remote processor, then it should work. > The second implementation uses request-based RMA and starts by calling MPI_Win_lock_all collectively on the window when it is created and MPI_Win_unlock_all when it is destroy so that the window is always in a passive synchronization epoch. The put is implement by calling MPI_Rput followed by calling MPI_Wait on the handle returned from the MPI_Rput call. Similarly, get is implemented by calling MPI_Rget followed by MPI_Wait. The wait call guarantees that the operation is completed locally and the data can then be used. However, from what I understand of the standard, it doesn’t say anything about the ordering of the operations so conceivably the put could execute remotely before the get. Inserting an MPI_Win_flush_all between the MPI_Rput and MPI_Rget should guarantee that the operations are ordered. > I’ve written the test program so that it can use either the lock or request-based implementations and I’ve also included an option that inserts a fence/flush plus barrier operation between put and get. The different configurations can be set up by defining some preprocessor symbols at the top of the program. The program loops over the test repeatedly and the current number of loops is set at 2000. The results I get running on a Linux cluster with an Infiniband network using OpenMPI-1.10.1 on 2 processors on 2 different SMP nodes are as follows: > Using OpenMPI-1.8.3: > Request-based implementation without synchronization: 9 successes out of 10 runs > Request-based implementation with synchronization: 19 successes out of 20 runs > Lock-based implementation without synchronization: 1 success out of 10 runs > Lock-based implementation with synchronization: 1 success out of 10 runs > Using OpenMPI-1.10.1 > Request-based implementation without synchronization: 2 successes out of 10 runs > Request-based implementation with synchronization: 8 successes out of 10 runs > Lock-based implementation without synchronization: 4 successes out of 10 runs > Lock-based implementation with synchronization: 2 successes out of 10 runs > Except for the request-based implementation without synchronization (in this case a call to MPI_Win_flush_all), I would expect these to all succeed. Is there some fault to my thinking here? I’ve attached the test program > Bruce Palmer > > > ___ > users mailing list > us...@open-mpi
Re: [OMPI users] Singleton process spawns additional thread
A singleton will indeed have an extra thread, but it should be quiescent. I’ll check the 1.10.2 release candidate and see if it still exhibits that behavior. > On Jan 7, 2016, at 9:32 PM, Au Eelis wrote: > > Hi! > > It is in so far related, that one of these threads is actually doing > something. > > Btw, I noticed this on two separate machines! A computing cluster with > admin-built openmpi and Archlinux with openmpi from the repositories. > > However, running the code with openmpi 1.6.2 and ifort 13.0.0 does not show > this behaviour. > > Best regards, > Stefan > > > > On 01/07/2016 03:27 PM, Sasso, John (GE Power, Non-GE) wrote: >> Stefan, I don't know if this is related to your issue, but FYI... >> >> >>> Those are async progress threads - they block unless something requires >>> doing >>> >>> On Apr 15, 2015, at 8:36 AM, Sasso, John (GE Power & Water, Non-GE) wrote: I stumbled upon something while using 'ps -eFL' to view threads of processes, and Google searches have failed to answer my question. This question holds for OpenMPI 1.6.x and even OpenMPI 1.4.x. >> >> For a program which is pure MPI (built and run using OpenMPI) and does not implement Pthreads or OpenMP, why is it that each MPI task appears as having 3 threads: >> >> UID PID PPID LWP C NLWPSZ RSS PSR STIME TTY TIME CMD sasso 20512 20493 20512 993 187849 582420 14 11:01 ? 00:26:37 /home/sasso/mpi_example.exe sasso 20512 20493 20588 03 187849 582420 11 11:01 ? 00:00:00 /home/sasso/mpi_example.exe sasso 20512 20493 20599 03 187849 582420 12 11:01 ? 00:00:00 /home/sasso/mpi_example.exe >> >> whereas if I compile and run a non-MPI program, 'ps -eFL' shows it running as a single thread? Granted the CPU utilization (C) for 2 of the 3 threads is zero, but the threads are bound to different processors (11,12,14). I am curious as to why this is, and no complaining that there is a problem. Thanks! >> >> --john >> >> >> -Original Message- >> From: users [mailto:users-boun...@open-mpi.org] On Behalf Of Au Eelis >> Sent: Thursday, January 07, 2016 7:10 AM >> To: us...@open-mpi.org >> Subject: [OMPI users] Singleton process spawns additional thread >> >> Hi! >> >> I have a weird problem with executing a singleton OpenMPI program, where an >> additional thread causes full load, while the master thread performs the >> actual calculations. >> >> In contrast, executing "mpirun -np 1 [executable]" performs the same >> calculation at the same speed but the additional thread is idling. >> >> In my understanding, both calculations should behave in the same way (i.e., >> one working thread) for a program which is simply moving some data around >> (mainly some MPI_BCAST and MPI_GATHER commands). >> >> I could observe this behaviour in OpenMPI 1.10.1 with ifort 16.0.1 and >> gfortran 5.3.0. I could create a minimal working example, which is appended >> to this mail. >> >> Am I missing something? >> >> Best regards, >> Stefan >> >> - >> >> MWE: Compile this with "mpifort main.f90". When executing with "./a.out", >> there is thread wasting cycles, while the master thread waits for input. >> When executing with "mpirun -np 1 ./a.out" this thread is idling. >> >> program main >> use mpi_f08 >> implicit none >> >> integer :: ierror,rank >> >> call MPI_Init(ierror) >> call MPI_Comm_Rank(MPI_Comm_World,rank,ierror) >> >> ! let master thread wait on [RETURN]-key >> if (rank == 0) then >> read(*,*) >> end if >> >> write(*,*) rank >> >> call mpi_barrier(mpi_comm_world, ierror) end program >> ___ >> users mailing list >> us...@open-mpi.org >> Subscription: >> https://urldefense.proofpoint.com/v2/url?u=http-3A__www.open-2Dmpi.org_mailman_listinfo.cgi_users&d=CwICAg&c=IV_clAzoPDE253xZdHuilRgztyh_RiV3wUrLrDQYWSI&r=tqKZ2vRCLufSSXPvzNxBrKr01YPimBPnb-JT-Js0Fmk&m=NPeEHKik35WrcHGDl5ZRq4IC6Le5g03o5YoqD9InrHw&s=eRYTNaknio7tNJFdOMTqvdlNNIq9p6evJoQxuvmqrLs&e= >> Link to this post: >> https://urldefense.proofpoint.com/v2/url?u=http-3A__www.open-2Dmpi.org_community_lists_users_2016_01_28237.php&d=CwICAg&c=IV_clAzoPDE253xZdHuilRgztyh_RiV3wUrLrDQYWSI&r=tqKZ2vRCLufSSXPvzNxBrKr01YPimBPnb-JT-Js0Fmk&m=NPeEHKik35WrcHGDl5ZRq4IC6Le5g03o5YoqD9InrHw&s=2_axdls1JH4Wm5MlkOXRrtXFb2LLVLCleKVx4ybpltU&e= >> ___ >> users mailing list >> us...@open-mpi.org >> Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users >> Link to this post: >> http://www.open-mpi.org/community/lists/users/2016/01/28238.php > > ___ > users mailing list > us...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users > Link to this post: > http://www.o
Re: [OMPI users] Singleton process spawns additional thread
Hi! It is in so far related, that one of these threads is actually doing something. Btw, I noticed this on two separate machines! A computing cluster with admin-built openmpi and Archlinux with openmpi from the repositories. However, running the code with openmpi 1.6.2 and ifort 13.0.0 does not show this behaviour. Best regards, Stefan On 01/07/2016 03:27 PM, Sasso, John (GE Power, Non-GE) wrote: Stefan, I don't know if this is related to your issue, but FYI... Those are async progress threads - they block unless something requires doing On Apr 15, 2015, at 8:36 AM, Sasso, John (GE Power & Water, Non-GE) wrote: I stumbled upon something while using 'ps -eFL' to view threads of processes, and Google searches have failed to answer my question. This question holds for OpenMPI 1.6.x and even OpenMPI 1.4.x. >> For a program which is pure MPI (built and run using OpenMPI) and does not implement Pthreads or OpenMP, why is it that each MPI task appears as having 3 threads: >> UID PID PPID LWP C NLWPSZ RSS PSR STIME TTY TIME CMD sasso 20512 20493 20512 993 187849 582420 14 11:01 ? 00:26:37 /home/sasso/mpi_example.exe sasso 20512 20493 20588 03 187849 582420 11 11:01 ? 00:00:00 /home/sasso/mpi_example.exe sasso 20512 20493 20599 03 187849 582420 12 11:01 ? 00:00:00 /home/sasso/mpi_example.exe >> whereas if I compile and run a non-MPI program, 'ps -eFL' shows it running as a single thread? Granted the CPU utilization (C) for 2 of the 3 threads is zero, but the threads are bound to different processors (11,12,14). I am curious as to why this is, and no complaining that there is a problem. Thanks! >> --john -Original Message- From: users [mailto:users-boun...@open-mpi.org] On Behalf Of Au Eelis Sent: Thursday, January 07, 2016 7:10 AM To: us...@open-mpi.org Subject: [OMPI users] Singleton process spawns additional thread Hi! I have a weird problem with executing a singleton OpenMPI program, where an additional thread causes full load, while the master thread performs the actual calculations. In contrast, executing "mpirun -np 1 [executable]" performs the same calculation at the same speed but the additional thread is idling. In my understanding, both calculations should behave in the same way (i.e., one working thread) for a program which is simply moving some data around (mainly some MPI_BCAST and MPI_GATHER commands). I could observe this behaviour in OpenMPI 1.10.1 with ifort 16.0.1 and gfortran 5.3.0. I could create a minimal working example, which is appended to this mail. Am I missing something? Best regards, Stefan - MWE: Compile this with "mpifort main.f90". When executing with "./a.out", there is thread wasting cycles, while the master thread waits for input. When executing with "mpirun -np 1 ./a.out" this thread is idling. program main use mpi_f08 implicit none integer :: ierror,rank call MPI_Init(ierror) call MPI_Comm_Rank(MPI_Comm_World,rank,ierror) ! let master thread wait on [RETURN]-key if (rank == 0) then read(*,*) end if write(*,*) rank call mpi_barrier(mpi_comm_world, ierror) end program ___ users mailing list us...@open-mpi.org Subscription: https://urldefense.proofpoint.com/v2/url?u=http-3A__www.open-2Dmpi.org_mailman_listinfo.cgi_users&d=CwICAg&c=IV_clAzoPDE253xZdHuilRgztyh_RiV3wUrLrDQYWSI&r=tqKZ2vRCLufSSXPvzNxBrKr01YPimBPnb-JT-Js0Fmk&m=NPeEHKik35WrcHGDl5ZRq4IC6Le5g03o5YoqD9InrHw&s=eRYTNaknio7tNJFdOMTqvdlNNIq9p6evJoQxuvmqrLs&e= Link to this post: https://urldefense.proofpoint.com/v2/url?u=http-3A__www.open-2Dmpi.org_community_lists_users_2016_01_28237.php&d=CwICAg&c=IV_clAzoPDE253xZdHuilRgztyh_RiV3wUrLrDQYWSI&r=tqKZ2vRCLufSSXPvzNxBrKr01YPimBPnb-JT-Js0Fmk&m=NPeEHKik35WrcHGDl5ZRq4IC6Le5g03o5YoqD9InrHw&s=2_axdls1JH4Wm5MlkOXRrtXFb2LLVLCleKVx4ybpltU&e= ___ users mailing list us...@open-mpi.org Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/users Link to this post: http://www.open-mpi.org/community/lists/users/2016/01/28238.php