[FLASH-USERS] Nei test unit in latest version
Gao-Yuan Zhang
gao-yuan.zhang at cfa.harvard.edu
Fri Feb 3 15:27:53 EST 2017
Hi all,
I am working on simulations using FLASH code. I found that the Nei test unit counters an error in FLASH 4.4. However, I successfully tested this unit in FLASH 4.2.2 with the same computer system and default settings.
The setup script is “./setup -auto neitest -1d”
All the files are remained default.
The intel fortran 2017 is used to compile and execute. The command line after “make” is
“mpirun -np 8 ./flash4”
using 8 CPUs.
The error messages are:
Element I.D.: 9 elsum = 0.99988912740675984
error in routine neimn: wrong sum of population fraction
DRIVER_ABORT: error in routine neimn: wrong sum of pop. fraction
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 4
Fatal error in PMPI_Reduce: Other MPI error, error stack:
PMPI_Reduce(1270).........: MPI_Reduce(sbuf=0x7fff2e29f610, rbuf=0x7fff2e29f650, count=7, MPI_DOUBLE_PRECISION, MPI_SUM, root=0, MPI_COMM_WORLD) failed
MPIR_Reduce_impl(1087)....:
MPIR_Reduce_intra(848)....:
MPIR_Reduce_impl(1087)....:
MPIR_Reduce_intra(895)....:
MPIR_Reduce_binomial(206).: Failure during collective
MPIR_Reduce_intra(828)....:
MPIR_Reduce_impl(1087)....:
MPIR_Reduce_intra(895)....:
MPIR_Reduce_binomial(144).:
dequeue_and_set_error(596): Communication error with rank 2
I have tried to change the accuracy in neimn routine to 0.1 rather than the default 1.e-4. The error exist anyway.
This may be a bug in the latest version. I have no idea what’s changed for the 4.4 version. Does any one know what’s happened with the nei unit? As far as I know the update of FLASH code doesn’t change this part.
Best wishes,
Gao-Yuan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20170203/ada064bc/attachment.htm>
More information about the flash-users
mailing list