Running SMAUG on Bede 1

Running the SMAUG Code on Bede

Introduction

The SMAUG project

What is Bede?

• Bede
• Bede has been Formally Handed Over
• Bede hardware

Documentation

• Bede Wiki
• Hackathon docs.

Installing Python Environment

Request interactive session

module load slurm
srun -A bdshe01 --pty bash

Steps which were taken

git clone https://github.com/smaug

From the source directory

make ot

• Modified makefile for Bede
• make_inputs_bede

Using the GPU hackathon docs

Main changes:

• The CUDA path CUDA = /opt/software/apps/libs/CUDA/10.1.243

• cuda FLAGS CUDACCFLAGS = –ptxas-options=-v -gencode=arch=compute_70,code=sm_70 -gencode=arch=compute_70,code=compute_70 -maxrregcount=32 -DUSE_SAC

Comment the code documentation needs much improvement! However, it wasn’t a major challenge to get the code to run on Bede. Next steps are as follows:

• Test jupyter notebook and plot outputs
• Check timings
• Check 3D codes