Skip to content

CRUNTCH4 Cluster at the CASCaM Center in UNT Chemistry

Remote Login

Connecting from Off Campus

First sign in to UNT VPN.

ssh <euid>@cruntch4.chem.unt.edu
openconnect: cisco-anyconnect alternative

Some linux users may encounter issues with cisco anyconnect. An alternative (some argue better than anyconnect) is openconnect.

For Ubuntu users (see this thread), do the following:

  • sudo apt-get install network-manager-openconnect-gnome
  • set openconnect protocol to "Cisco AnyConnect"
  • set openconnect gateway to "vpn.unt.edu"
  • start connection by typing UNT credentials (EUID and password)

System-Specific Settings

example submission scripts

Samples can be found at /storage/nas_scr/shared/slurm_scripts/

filesystems
  • Home: $HOME
  • Network Attached Storage (NAS) scratch: /storage/nas_scr/$USER
  • Archive storage: /archive/$USER
make a link to the scratch partition 
cd ~ # go home
ln -s /storage/nas_scr/$USER scr  # make a link named scr at home that points to the NAS scratch where calculations happen

Compilation of Quantum ESPRESSO

#!/bin/bash
module load INTEL/compiler-rt/2022.2.1 INTEL/compiler/2022.2.1
module load INTEL/mkl/2022.2.1
module load INTEL/mpi/2021.7.1
module load INTEL/tbb/2021.7.1   
module load INTEL/oclfpga/2022.2.1
module load git/2.33.1

WRK_DIR=build.cruntch4.cpu

cmake  -DCMAKE_CXX_COMPILER=icpc -DCMAKE_C_COMPILER=icc -DCMAKE_Fortran_COMPILER=mpiifort \
-DQE_ENABLE_OPENMP=ON -DCMAKE_Fortran_FLAGS='-O3 -g -cpp -qopenmp -qopenmp-simd' -B $WRK_DIR

cd $WRK_DIR
make -j pw

#!/bin/bash

module load INTEL/compiler/2022.2.1
module load INTEL/mkl/2022.2.1
module load INTEL/mpi/2021.7.1
module load cuda11.7/toolkit/11.7.1
module load nvhpc/22.7
module load git/2.33.1

WRK_DIR=build.cruntch4.gpu

cmake -DQE_ENABLE_OPENMP=ON -DQE_ENABLE_CUDA=ON -DQE_ENABLE_MPI=OFF  -DQE_ENABLE_OPENACC=ON \
-DQE_ENABLE_PROFILE_NVTX=ON -DCMAKE_Fortran_FLAGS='-O3 -g -fast -acc -gpu=cc80 -Mcache_align -Mpreprocess -Mlarge_arrays -gpu=managed' -B $WRK_DIR

cd $WRK_DIR
make -j 32 pw

#!/bin/bash

module load INTEL/mkl/2022.2.1
module load OPENMPI/gcc12.2/cuda11.7.1/4.1.4
module load cuda11.7/toolkit/11.7.1
module load nvhpc/22.7
module load git/2.33.1

WRK_DIR=build.cruntch4.gpu.ompi

cmake -DQE_ENABLE_OPENMP=ON -DQE_ENABLE_CUDA=ON -DQE_ENABLE_MPI=OFF  -DQE_ENABLE_OPENACC=ON \
-DQE_ENABLE_PROFILE_NVTX=ON -DCMAKE_Fortran_FLAGS='-O3 -g -fast -acc -gpu=cc80 -Mcache_align -Mpreprocess -Mlarge_arrays -gpu=managed' -B $WRK_DIR

cd $WRK_DIR
make -j 32 pw