Hello All,
I am consistingly finding that the binary write operations are quite faster than
HDF5. So I am actually doubting the way I measure the performance.
I am new to parallel programming so kindly excuse if this sounds lame. This is
how measure the PHDF5 write time in my code:
do k1 = 1, 15
if(iproc.lt.nend) then
start = MPI_WTIME()
CALL phdfwrite(pdsetname,pdset_id,pdataspace,
& vampnew,i1reach,parallelOffset,1,0)
end = mpi_wtime()
CALL MPI_Barrier(new_comm,error)
CALL MPI_allreduce(end,endall,1,MPI_REAL8,MPI_MAX,new_comm,
& error)
call MPI_Allreduce(start,startall,1,MPI_REAL8,MPI_MIN,
& new_comm,error)
endif
enddo
writetime = endall - startall
And I measure the binary write as follows:
do k1 = 1, 15
if (iproc.eq.0) then
start0 = MPI_WTIME()
write(17) vampnew(1:i1reach)
end0 = MPI_WTIME()
endif
if (nproc.gt.1) then
do kk = 2, 15
if (iproc.eq.kk-1) call MPI_SEND(samp)
if (iproc.eq.0) then
call MPI_RECV(samp)
if(kk.eq.2) start1 = MPI_WTIME()
write(17) samp(1:ileng)
if(kk.eq.15) end1 = MPI_WTIME()
endif
enddo
endif
enddo
writetime = (end0-start0)+(end1-start1)
Is there something wrong I am doing that someone can point out please ?
Thanks a lot.
Regards,
Nikhil
···
----------------------------------------------------------------------
This mailing list is for HDF software users discussion.
To subscribe to this list, send a message to hdf-forum-subscribe@hdfgroup.org.
To unsubscribe, send a message to hdf-forum-unsubscribe@hdfgroup.org.