MSSM_HWW.submit

#!/usr/bin/env zsh

### Give the job a name, will be displayed in bjobs  default: Job_with_no_name
#BSUB -J test_theano_env_setup.job

### Max execution time in   hour:minutes   default: 00:15
#BSUB -W 24:00

### Max memory per core (RAM) in MB default: 512
### the product of numbers of cores and max memory has to be < 126000
#BSUB -M 5000

### number of cores [1-24] default: 1
#BSUB -n 12
# give a path for log and error file. %J is Job number %I is Job index
#BSUB -o /home/mf278754/master/output/out.out
#BSUB -e /home/mf278754/master/output/out.err

### these options are needed to run on our mashine
### use -a gpu if -n = 1 or default
### use -a "gpu openmp" if -n > 1
#BSUB -a "gpu openmp"
#BSUB -R pascal
#BSUB -P phys3b

### set up stuff

##source /cvmfs/sft.cern.ch/lcg/views/LCG_91/x86_64-slc6-gcc62-opt/setup.sh
##. /cvmfs/sft.cern.ch/lcg/app/releases/ROOT/6.10.08/x86_64-centos7-gcc48-opt/root/bin/thisroot.sh
source /home/phys3b/Envs/keras_tf_sharedUsers/bin/activate
##source /cvmfs/sft.cern.ch/lcg/views/LCG_91/x86_64-slc6-gcc62-opt/setup.sh
##. /cvmfs/sft.cern.ch/lcg/app/releases/ROOT/6.08.06/x86_64-centos7-gcc48-opt/root/bin/thisroot.sh

# If you want to use your own .theanorc
#export THEANORC=~/.theanorc

export CUDA_VISIBLE_DEVICES=`/home/phys3b/etc/check_gpu.py  2`
if [ '$CUDA_VISIBLE_DEVICES' = '-1' ];
then
echo '##### GPUs busy. Restart job later.' exit 1
else
echo 'Found free GPU devices :'
echo 'CUDA_VISIBLE_DEVICES =  $CUDA_VISIBLE_DEVICES '
fi

##echo ""
##echo "\n\nDeviceQuery:\n"
##/home/phys3b/test/test_cuda_examples/NVIDIA_CUDA-8.0_Samples/bin/x86_64/linux/release/deviceQuery
echo "\n\nnvidia-smi\n"
nvidia-smi

#echo "\n\nyour pygpu installation:"


#echo "\n\nTest pygpu\n"
#export DEVICE="cuda1"

#export CUDA_VISIBLE_DEVICES="0"

python /home/mf278754/master/Machine-Learning/training/train.py /home/mf278754/master/Machine-Learning/config/MSSM_HWW.yaml