Hi,
I just compiled and installed MEEP 1.1.1 on BG/L (w/ parallel HDF5, w/o
libctl or Guile).
The 'symmetry' test fails, and all 15 other tests pass fine. Is this a
real thing? This looks to me like a too restrictive testing criteria and
not a true failure? Please see below and the attached file for details.
Thanks,
Seppo
XXXXXXXXXXXXXXXXXXXXX
...
Testing X twofold rotational symmetry with PML...
real part = 4.79458e-08 differs by -1.10102e-16 from 4.79458e-08
This gives a fractional error of 2.29639e-09
hy differs by 1.10102e-16+3.2831e-18i from 4.79458e-08-6.15267e-06i
This comes out to a fractional error of 1.79024e-11
Right now I'm looking at x = 1.46, y = 0.33, time 29.4375
meep: meep: meep: meep: meep: meep: error in exact_pml_rot2x_tm vacuum
...
XXXXXXXXXXXXXXXXXXXXXX
Configuration:
export PATH=$PATH
export LD_LIBRARY_PATH=/opt/ibmmath/lib/:$LD_LIBRARY_PATH
export CC=xlc
export CXX=xlC
export FC=xlf90
export MPICC=mpixlc
export MPICXX=mpixlcxx
export MPIFC=mpixlf90
#export CC=
#export CXX=
#export FC=
#export MPICC=mpicc
#export MPICXX=mpicxx
#export MPIFC=mpif77
export CFLAGS="-O2 -I/bgl/BlueLight/ppcfloor/bglsys/include
-DMPICH_IGNORE_CXX_SEEK -qarch=440d -qtune=440 "
#export CFLAGS="-O2 "
export CPPFLAGS=$CFLAGS
export LDFLAGS="-L/bgl/BlueLight/ppcfloor/bglsys/lib"
#export LIBS=" -lmpich.rts -lcxxmpich.rts -lmsglayer.rts -lrts.rts
-ldevices.rts -lm -lctl -lc -lnss_files -lnss_dns -lresolv "
export LIBS=" -lmpich.rts -lcxxmpich.rts -lmsglayer.rts -lrts.rts
-ldevices.rts "
#export MPILIBS=$LIBS
export MAKE=gmake
# parallel MEEP, w/ parallel HDF5
./configure --prefix=/bghome/meep/meep_parallelhdf5 --with-mpi
--with-blas=-lesslbg --with-lapack=-lesslbg
--with-hdf5=/bghome/meep/hdf5_parallel --without-libctl
XXXXXXXXXXXXXXXXXXXXXX
--
Seppo Sahrakorpi, Ph.D.
Senior Research Engineer, User Services Lead
Instructional and Research Computing Services
Harvard University School of Engineering and Applied Sciences IT
Maxwell Dworkin MD211, 33 Oxford St., Cambridge, MA 02138
Phone: 617-495-3411 Email: [email protected]
======================================
meep 1.1.1: tests/test-suite.log
======================================
1 of 16 tests failed.
.. contents:: :depth: 2
FAIL: symmetry (exit: 1)
========================
<Aug 24 20:32:14.959086> FE_MPI (Info) : Initializing MPIRUN
<Aug 24 20:32:14.987115> FE_MPI (Info) : Scheduler interface library loaded
<Aug 24 20:32:15.118189> LoadLeveler (Info) : get_parameters() - Passing
jobname=bgfen.localnet.16701.0 and partition_id=RMP24Au202345439 to mpirun.
<Aug 24 20:32:15.380231> BRIDGE (Info) : The machine serial number (alias) is
BGL
<Aug 24 20:32:15.521224> FE_MPI (Info) : Preparing partition
<Aug 24 20:32:15.521975> BE_MPI (Info) : Examining specified partition
<Aug 24 20:32:15.698434> BE_MPI (Info) : Checking partition RMP24Au202345439
initial state ...
<Aug 24 20:32:15.698476> BE_MPI (Info) : Partition RMP24Au202345439 initial
state = READY ('I')
<Aug 24 20:32:15.698494> BE_MPI (Info) : Checking partition owner...
<Aug 24 20:32:15.698515> BE_MPI (Info) : Checking if the partition is busy ...
<Aug 24 20:32:15.988349> BE_MPI (Info) : Checking partition size ...
<Aug 24 20:32:15.988385> BE_MPI (Info) : Partition owner (root) differs from
the current user (seppo)
<Aug 24 20:32:15.988400> BE_MPI (Info) : Checking if the user is in the
partition users list...
<Aug 24 20:32:15.988415> BE_MPI (Info) : Current user is in the partition user
list
<Aug 24 20:32:16.089523> BE_MPI (Info) : Done preparing partition
<Aug 24 20:32:16.129845> FE_MPI (Info) : Adding job
<Aug 24 20:32:16.719208> FE_MPI (Info) : Job added with the following id: 17322
<Aug 24 20:32:16.719242> FE_MPI (Info) : Starting job 17322
<Aug 24 20:32:16.901723> FE_MPI (Info) : Waiting for job to terminate
<Aug 24 20:32:22.415633> BE_MPI (Info) : IO - Connection from Ciodb established
on fd=12
<Aug 24 20:32:22.418569> BE_MPI (Info) : IO - Threads initialized
Using MPI version 2.0, 8 processes
Testing with various kinds of symmetry...
Testing Z mirror symmetry in 1D...
Testing Z mirror symmetry in Cylindrical...
Testing Y periodic with mirror symmetry...
Testing Y periodic with mirror symmetry...
Testing two mirrors with PML...
Testing origin shift in 2D...
Testing X twofold rotational symmetry with PML...
real part = 4.79458e-08 differs by -1.10102e-16 from 4.79458e-08
This gives a fractional error of 2.29639e-09
hy differs by 1.10102e-16+3.2831e-18i from 4.79458e-08-6.15267e-06i
This comes out to a fractional error of 1.79024e-11
Right now I'm looking at x = 1.46, y = 0.33, time 29.4375
meep: meep: meep: meep: meep: meep: error in exact_pml_rot2x_tm vacuum
meep: error in exact_pml_rot2x_tm vacuum
error in exact_pml_rot2x_tm vacuum
error in exact_pml_rot2x_tm vacuum
error in exact_pml_rot2x_tm vacuum
error in exact_pml_rot2x_tm vacuum
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 5error in
exact_pml_rot2x_tm vacuum
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 7application called
MPI_Abort(MPI_COMM_WORLD, 1) - process 2application called
MPI_Abort(MPI_COMM_WORLD, 1) - process 1application called
MPI_Abort(MPI_COMM_WORLD, 1) - process 4application called
MPI_Abort(MPI_COMM_WORLD, 1) - process 6meep: application called
MPI_Abort(MPI_COMM_WORLD, 1) - process 3error in exact_pml_rot2x_tm vacuum
application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0<Aug 24
20:32:32.635017> BE_MPI (Info) : Job 17322 switched to state TERMINATED ('T')
<Aug 24 20:32:32.635064> BE_MPI (Info) : Job successfully terminated
<Aug 24 20:32:32.840720> BE_MPI (ERROR): The error message in the job record is
as follows:
<Aug 24 20:32:32.840755> BE_MPI (ERROR): "killed by exit(1) on node 14"
<Aug 24 20:32:32.981657> FE_MPI (Info) : BG/L job exit status = (1)
<Aug 24 20:32:32.981700> FE_MPI (Info) : Job terminated normally
<Aug 24 20:32:32.982307> BE_MPI (Info) : Starting cleanup sequence
<Aug 24 20:32:32.982336> BE_MPI (Info) : BG/L Job already terminated / hasn't
been added
<Aug 24 20:32:32.984827> BE_MPI (ERROR): The error message in the job record is
as follows:
<Aug 24 20:32:32.984857> BE_MPI (ERROR): "killed by exit(1) on node 14"
<Aug 24 20:32:32.984876> BE_MPI (Info) : Partition was supplied with READY
('I') initial state
<Aug 24 20:32:32.984891> BE_MPI (Info) : No need to destroy the partition
<Aug 24 20:32:33.196745> BE_MPI (Info) : == BE completed ==
<Aug 24 20:32:33.202258> FE_MPI (Info) : == FE completed ==
<Aug 24 20:32:33.202295> FE_MPI (Info) : == Exit status: 1 ==
<Aug 24 20:32:33.202335> LoadLeveler (Info) : mpirun_done() - mpirun_rc=1
_______________________________________________
meep-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/meep-discuss
