! Copyright (c) 2017 Jiaxing Qi <jiaxing.qi@uni-siegen.de> ! ! Redistribution and use in source and binary forms, with or without ! modification, are permitted provided that the following conditions are met: ! ! 1. Redistributions of source code must retain the above copyright notice, this ! list of conditions and the following disclaimer. ! ! 2. Redistributions in binary form must reproduce the above copyright notice, ! this list of conditions and the following disclaimer in the documentation ! and/or other materials provided with the distribution. ! ! THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" ! AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE ! IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE ! DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE ! FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL ! DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR ! SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER ! CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, ! OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE ! OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. !mpi!nprocs = 4 program sparse_alltoall_test use mpi use tem_sparse_comm_module, only: tem_sparse_alltoall_int implicit none ! Assume we have 4 procs: 0 - 1 - 2 - 3 ! Each proc send data to their direct neighbors ! the data is a single integer whose value is the targeting proc, i.e. ! proc 0 sends value 1 to proc 1 ! proc 1 sends value 0 to proc 0, value 2 to proc 2 ! proc 2 sends value 1 to proc 1, value 3 to proc 3 ! proc 3 sends value 2 to proc 2 ! After communication, each proc should have the following data: ! proc 0: 0 ! proc 1: 1 1 ! proc 2: 2 2 ! proc 3: 3 integer :: iError integer :: myrank, nProcs logical :: local_success, success ! integer :: rstat(MPI_STATUS_SIZE) integer, allocatable :: targets(:) integer, allocatable :: send_buffer(:) integer, allocatable :: sources(:) integer, allocatable :: recv_buffer(:) call MPI_Init(iError) call MPI_Comm_rank(MPI_COMM_WORLD, myrank, iError) call MPI_COMM_SIZE(MPI_COMM_WORLD, nProcs, IERROR) if ( myrank == 0 ) then ! has one neighbor allocate( targets(1) ) allocate( send_buffer(1) ) targets(1) = myrank + 1 send_buffer(1) = myrank + 1 else if ( myrank == (nProcs-1) ) then ! has one neighbor allocate( targets(1) ) allocate( send_buffer(1) ) targets(1) = myrank - 1 send_buffer(1) = myrank - 1 else ! has two neighbor allocate( targets(2) ) allocate( send_buffer(2) ) targets(1) = myrank - 1 targets(2) = myrank + 1 send_buffer(1) = myrank - 1 send_buffer(2) = myrank + 1 end if call tem_sparse_alltoall_int( targets, send_buffer, & & sources, recv_buffer, MPI_COMM_WORLD) if ( myrank == 0 .or. myrank == (nProcs-1) ) then local_success = (recv_buffer(1) == myrank) else local_success = ((recv_buffer(1) == myrank) .and. (recv_buffer(2) == myrank)) end if if ( .not. local_success ) then write(*,*) "Rank: ", myrank, " sources: ", sources write(*,*) "Rank: ", myrank, " recv_bu: ", recv_buffer end if call MPI_REDUCE(local_success, success, 1, MPI_LOGICAL, MPI_LAND, 0, MPI_COMM_world, IERROR) if ( myrank == 0 ) then if ( success ) then write(*,*) "PASSED" else write(*,*) "FAILED" end if end if call MPI_Finalize(iError) end program sparse_alltoall_test