File tree Expand file tree Collapse file tree 2 files changed +31
-6
lines changed Expand file tree Collapse file tree 2 files changed +31
-6
lines changed Original file line number Diff line number Diff line change @@ -102,10 +102,36 @@ function module_load_string(::Type{DerechoBackend})
102102end
103103
104104function module_load_string (:: Type{GCPBackend} )
105- return """ export OPAL_PREFIX="/sw/openmpi-5.0.5"
106- export PATH="/sw/openmpi-5.0.5/bin:\$ PATH"
107- export LD_LIBRARY_PATH="/sw/openmpi-5.0.5/lib:\$ LD_LIBRARY_PATH"
108- export UCX_MEMTYPE_CACHE=y # UCX Memory optimization which toggles whether UCX library intercepts cu*alloc* calls
105+ return """
106+ unset CUDA_ROOT
107+ unset NVHPC_CUDA_HOME
108+ unset CUDA_INC_DIR
109+ unset CPATH
110+ unset NVHPC_ROOT
111+
112+ # NVHPC and HPC-X paths
113+ export NVHPC=/sw/nvhpc/Linux_x86_64/24.5
114+ export HPCX_PATH=${NVHPC}/comm_libs/12.4/hpcx/hpcx-2.19
115+
116+ # CUDA environment
117+ export CUDA_HOME=${NVHPC}/cuda/12.4
118+ export CUDA_PATH=${CUDA_HOME}
119+ export CUDA_ROOT=${CUDA_HOME}
120+
121+ # MPI via MPIwrapper
122+ export MPITRAMPOLINE_LIB="/sw/mpiwrapper/lib/libmpiwrapper.so"
123+ export OPAL_PREFIX=${HPCX_PATH}/ompi
124+
125+ # Library paths - CUDA first, then HPC-X
126+ export LD_LIBRARY_PATH="${CUDA_HOME}/lib64:${HPCX_PATH}/ompi/lib${LD_LIBRARY_PATH:+:$LD_LIBRARY_PATH }"
127+
128+ # Executable paths
129+ export PATH=/sw/mpiwrapper/bin:${CUDA_HOME}/bin:${PATH}
130+ export PATH="${NVHPC}/profilers/Nsight_Systems/target-linux-x64:${PATH}"
131+
132+ # Julia
133+ export PATH="/sw/julia/julia-1.11.5/bin:${PATH}"
134+ export JULIA_MPI_HAS_CUDA=true
109135 """
110136end
111137
Original file line number Diff line number Diff line change @@ -212,8 +212,7 @@ function generate_sbatch_script(
212212 climacomms_device = gpus_per_task > 0 ? " CUDA" : " CPU"
213213 # TODO : Remove this exception for GCP
214214 mpiexec_string =
215- get_backend () == GCPBackend ?
216- " /sw/openmpi-5.0.5/bin/mpiexec -n $ntasks " :
215+ get_backend () == GCPBackend ? " mpiexec -n $ntasks " :
217216 " srun --output=$member_log --open-mode=append"
218217 sbatch_contents = """
219218 #!/bin/bash
You can’t perform that action at this time.
0 commit comments