From 698af4b24801c3f061692f255aec5a5d4c8323b3 Mon Sep 17 00:00:00 2001
From: allaffa <massimiliano.lupo.pasini@gmail.com>
Date: Sat, 28 Sep 2024 00:12:59 -0400
Subject: [PATCH] remove job submission script previously added by mistake

---
 job-frontier-ogb-deepspeed.sh | 46 --------------------------------
 job-frontier-preonly-nvme.sh  | 50 -----------------------------------
 2 files changed, 96 deletions(-)
 delete mode 100644 job-frontier-ogb-deepspeed.sh
 delete mode 100755 job-frontier-preonly-nvme.sh

diff --git a/job-frontier-ogb-deepspeed.sh b/job-frontier-ogb-deepspeed.sh
deleted file mode 100644
index 4b464ae13..000000000
--- a/job-frontier-ogb-deepspeed.sh
+++ /dev/null
@@ -1,46 +0,0 @@
-#!/bin/bash
-#SBATCH -A LRN031
-#SBATCH -J HydraGNN
-#SBATCH -o job-%j.out
-#SBATCH -e job-%j.out
-#SBATCH -t 01:00:00
-#SBATCH -p batch
-#SBATCH -q debug
-#SBATCH -N 8
-#SBATCH -S 1
-
-ulimit -n 65536
-
-export MPICH_ENV_DISPLAY=0
-export MPICH_VERSION_DISPLAY=0
-export MIOPEN_DISABLE_CACHE=1
-export NCCL_PROTO=Simple
-
-export OMP_NUM_THREADS=7
-export HYDRAGNN_NUM_WORKERS=0
-export HYDRAGNN_USE_VARIABLE_GRAPH_SIZE=1
-export HYDRAGNN_AGGR_BACKEND=mpi
-export HYDRAGNN_VALTEST=0 
-export NCCL_P2P_LEVEL=NVL
-export NCCL_P2P_DISABLE=1
-
-export MPICH_GPU_SUPPORT_ENABLED=1
-export MPICH_GPU_MANAGED_MEMORY_SUPPORT_ENABLED=1
-export MPICH_OFI_NIC_POLICY=GPU
-
-
-source /lustre/orion/cph161/world-shared/mlupopa/module-to-load-frontier.sh
-
-source /lustre/orion/cph161/world-shared/mlupopa/max_conda_envs_frontier/bin/activate
-conda activate hydragnn
-
-export PYTHONPATH=/lustre/orion/cph161/world-shared/mlupopa/ADIOS_frontier/install/lib/python3.8/site-packages/:$PYTHONPATH
-
-export PYTHONPATH=$PWD:$PYTHONPATH
-
-
-# both commands should work
-srun -N$SLURM_JOB_NUM_NODES -n$((SLURM_JOB_NUM_NODES*8)) -c7 --gres=gpu:8 \
-   python -u ./examples/ogb/train_gap.py gap --adios --use_deepspeed
-# srun -N$SLURM_JOB_NUM_NODES -n$((SLURM_JOB_NUM_NODES*8)) -c7 --gpus-per-task=1 --gpu-bind=closest \
-#     python -u ./examples/ogb/train_gap.py gap --adios --use_deepspeed
diff --git a/job-frontier-preonly-nvme.sh b/job-frontier-preonly-nvme.sh
deleted file mode 100755
index c9085d225..000000000
--- a/job-frontier-preonly-nvme.sh
+++ /dev/null
@@ -1,50 +0,0 @@
-#!/bin/bash
-#SBATCH -A CPH161
-#SBATCH -J HydraGNN
-#SBATCH -o job-%j.out
-#SBATCH -e job-%j.out
-#SBATCH -t 02:00:00
-#SBATCH -p batch
-#SBATCH -N 5
-#SBATCH -q debug
-#SBATCH -S 0
-#SBATCH -C nvme
-
-export MPICH_ENV_DISPLAY=1
-export MPICH_VERSION_DISPLAY=1
-export MPICH_GPU_SUPPORT_ENABLED=1
-export MPICH_GPU_MANAGED_MEMORY_SUPPORT_ENABLED=1
-export MPICH_OFI_NIC_POLICY=GPU
-export MIOPEN_DISABLE_CACHE=1
-export NCCL_PROTO=Simple
-
-export OMP_NUM_THREADS=7
-export HYDRAGNN_AGGR_BACKEND=mpi
-
-source /lustre/orion/cph161/world-shared/mlupopa/module-to-load-frontier.sh
-
-source /lustre/orion/cph161/world-shared/mlupopa/max_conda_envs_frontier/bin/activate
-conda activate hydragnn
-
-export PYTHONPATH=/lustre/orion/cph161/world-shared/mlupopa/ADIOS_frontier/install/lib/python3.8/site-packages/:$PYTHONPATH
-
-export PYTHONPATH=$PWD:$PYTHONPATH
-cd  examples/mptrj/
-
-# SBCAST file from Orion to NVMe -- NOTE: ``-C nvme`` is required to use the NVMe drive
-sbcast -pf dataset/MPtrj_2022.9_full.json /mnt/bb/$USER/MPtrj_2022.9_full.json
-if [ ! "$?" == "0" ]; then
-    # CHECK EXIT CODE. When SBCAST fails, it may leave partial files on the compute nodes, and if you continue to launch srun,
-    # your application may pick up partially complete shared library files, which would give you confusing errors.
-    echo "SBCAST failed!"
-    exit 1
-fi
-
-echo
-# Showing the file on the current node -- this will be the same on all other nodes in the allocation
-echo "*****SBCAST FILE ON CURRENT NODE******"
-ls /mnt/bb/$USER/
-echo "**************************************"
-
-srun -n$((SLURM_JOB_NUM_NODES*4)) python -u train.py --preonly --pickle --tmpfs "/mnt/bb/$USER/" 
-#python -u train.py --preonly