On Sun, Feb 8, 2015 at 11:43 AM, Gideon Simpson <[email protected]> wrote:
> What should I make of the following warning, where I have used an > MPI_Allreduce on a quantity computed on each processor. > > include -I/opt/local/include -I/opt/local/include/mpich-gcc48 > -I/opt/local/include `pwd`/petsc_trapz1.c > In file included from /opt/local/lib/petsc/include/petscsys.h:1794:0, > from /opt/local/lib/petsc/include/petscis.h:7, > from /opt/local/lib/petsc/include/petscvec.h:9, > from /Users/gideon/code/trapz/petsc_trapz1.c:3: > /Users/gideon/code/trapz/petsc_trapz1.c: In function 'main': > /opt/local/lib/petsc/include/petsclog.h:370:57: warning: value computed is > not used [-Wunused-value] > ((petsc_allreduce_ct += PetscMPIParallelComm(comm),0) || > MPI_Allreduce(sendbuf,recvbuf,count,datatype,op,comm)) > ^ > /Users/gideon/code/trapz/petsc_trapz1.c:110:3: note: in expansion of macro > 'MPI_Allreduce' > MPI_Allreduce(&local_trapz, &global_trapz, 1, MPI_DOUBLE, MPI_SUM, > PETSC_COMM_WORLD); > It looks like you are not checking the return code: ierr = MPI_Allreduce(&local_trapz, &global_trapz, 1, MPI_DOUBLE, MPI_SUM, PETSC_COMM_WORLD);CHKERRQ(ierr); Matt > > -gideon > > -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener
