Revision as of 10:30, 16 December 2021

Welcome to the Fisher Group Wiki!

Wiki Organization

This top-level page structures the Wiki into major topics of current research within the group. Please do not add to this page directly, but instead drill down to the next level to begin editing your area of research. To learn more about editing pages, please see the page About the Fisher Group Wiki.

@@ Line 89: / Line 89: @@
 ===[[FLASHXnet on Expanse]]===
-[A] On the login node:
-. Run these commands:
-module reset
-conda activate py2
-here py2 is a python2 environment which one needs to set up first for the setup.py to work. We need to re-write the setup.py in python3 in near future.
-. After that, setup the model by:
- ./setup Hybrid_WDs_scatter -objdir=D6_test_scatter -3d -auto +cube32 -maxblocks=30 +uhd +newMpole +hdf5typeIO withParticles=False xnet=True xnetGPU=True xnetData=Data_SN55 -opt
-Then enter an interactive job on a GPU node by running the following command:
- srun --partition=gpu-debug --pty --account=TG-AST100038 --ntasks-per-node=40 --nodes=1 --mem=356G --gpus=1 -t 0:30:00 --wait=0 --export=ALL /bin/bash
-This will request an interactive session on 1 node with 40 cores, 1 GPU, 356GB of memory for 30 minutes.
-[B] On the GPU node (after srun):
-. Load the following modules:
- conda activate py2
- module reset
- module load subversion
- module load intel
- module load openmpi
- module load hdf5/1.8.21
- module load cuda10.2/toolkit
-. Enter the build directory and then perform:
- make -j4
 ===[[FLASH - TORCH - SUPERNU - SNID Pipeline]]===

Main Page: Difference between revisions

Revision as of 10:30, 16 December 2021

Wiki Organization

Navigation menu

Search