I have just configured and compiled abinit 7.6.4 with the following .ac config:
Code: Select all
enable_64bit_flags="yes"
prefix="${HOME}/software/abinit"
CPP="icc -E"
enable_mpi="yes"
enable_mpi_io="yes"
with_mpi_prefix="/usr/local"
enable_gpu="yes"
with_gpu_flavor="cuda-double"
with_gpu_prefix="/usr/local/cuda-5.5"
with_trio_flavor="netcdf+etsf_io+fox"
with_fft_flavor="fftw3-mkl"
with_fft_libs="-L/opt/intel/composer_xe_2013_sp1.2.144/mkl/lib/intel64 -Wl,--start-group -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -Wl,--end-group"
with_linalg_flavor="mkl+magma"
with_linalg_libs="-L/usr/local/magma/lib -lmagma -lmagmablas -L/opt/intel/composer_xe_2013_sp1.2.144/mkl/lib/intel64 -Wl,--start-group -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -Wl,--end-group"
with_dft_flavor="atompaw+bigdft+libxc+wannier90"
enable_gw_dpc="yes"
enable_test_timeout="yes"
magma libs were generated with this make.inc file:
Code: Select all
GPU_TARGET = Tesla
CC = icc
NVCC = nvcc
FORT = ifort
ARCH = ar
ARCHFLAGS = cr
RANLIB = ranlib
OPTS = -O3 -DADD_ -Wall -openmp -DMAGMA_WITH_MKL -DMAGMA_SETAFFINITY
F77OPTS = -O3 -DADD_ -warn all
FOPTS = -O3 -DADD_ -warn all
NVOPTS = -O3 -DADD_ -Xcompiler -fno-strict-aliasing
LDOPTS = -openmp
LIB = -lmkl_intel_lp64 -lmkl_intel_thread -lmkl_core -lpthread -lcublas -lcudart -lstdc++ -lm
-include make.check-mkl
-include make.check-cuda
LIBDIR = -L$(MKLROOT)/lib/intel64 \
-L$(CUDADIR)/lib64
INC = -I$(CUDADIR)/include -I$(MKLROOT)/include
On the other hand, the OpenMPI environment was compiled using the Intel 14 compilers
The ./configure went fine, the make also did... but, when running the runtests.py I received a lot of errors mainly divided in two categories:
1. EXAMPLE:
[tutorespfn][telast_5] fldiff.pl fatal error:
The diff analysis cannot be pursued : the leading characters differ.
File .../abinit-7.6.4/tests/tutorespfn/Refs/telast_5.out, line 1206, 144 ignored, character:=
File .../abinit-7.6.4/build/temp_test/Test_suite/tutorespfn_telast_5/telast_5.out, line 1186, 124 ignored, character:
when eye-examined the output files, it looks like a correct output... and when the values are compared, nothing looks so tremendous. could you advice me with the possible origin of this error? must be taken very seriously or is just a bug in any of the sed-awk-diff routines?
2. EXAMPLE:
[tutorial][tbase2_5] failed: absolute error 9.132 > 3.681e-09
this deserves no comments... any good advice??
Best regards, boys... and keep the good work!
