Working with SCC (Interactive, Jupyter and VS Code)

Interactive Jobs

Single GPU

qrsh -P batmanlab -l h_rt=13:00:00 -pe omp 8 -l mem_per_core=4G -l gpus=1 -l gpu_c=6.0

• -P: Project
• h_rt: #hours you want to use
• gpu_c: Which particular gpu you want, usually 6.0, 7.0, 7.5, 8.0, 8.5

Also after you are allocated with a GPU, type these 3 command in the following order:

module load miniconda/23.1.0
module load python3/3.8
conda activate /restricted/projectnb/batmanlab/shawn24/breast_clip_rtx_6000

The last one must be your conda environment, not mine

Multiple GPU

qrsh -P batmanlab -l h_rt=13:00:00 -pe omp 8 -l mem_per_core=4G -l gpus=4 -l gpu_c=7.0

Check all the jobs submitted by you

qstat -u shawn24

All available GPUS

qgpus -V

BatmanLab path

/restricted/projectnb/batmanlab Create a folder with your BuID in this directory.

Datasets:

/restricted/projectnb/batmanlab/shared/Data/

Running Jupyter

Go to scc-ondemand.bu.edu and follow the below setting (change Working Directory to your path):

Running VScode

Go to scc-ondemand.bu.edu and follow the below setting (change Working Directory to your path):

You can directly code in VS Code server.