Merge pull request #39 from carlos-encs/jupyter

Jupyter: tunnel and examples - updated

doc/web/speed-manual.css

@@ -62,11 +62,11 @@ div.par-math-display, div.math-display{text-align:center;}
 body{ margin:1em auto; max-width:80ch; padding:0 .62em; }
 h1,h2,h3,h4,h5 { line-height:1.2; }
 @media print{ body{ max-width:none } }
-.partHead, likepartHead { font-size: 2em; }
-.chapterHead, likechapterHead { font-size: 1.7411em; }
-.sectionHead, likesectionHead { font-size: 1.5157em; }
-.subsectionHead, likesubsectionHead { font-size: 1.3195em; }
-.subsubsectionHead, likesubsubsectionHead { font-size: 1.1487em; }
+.partHead, .likepartHead { font-size: 2em; }
+.chapterHead, .likechapterHead { font-size: 1.7411em; }
+.sectionHead, .likesectionHead { font-size: 1.5157em; }
+.subsectionHead, .likesubsectionHead { font-size: 1.3195em; }
+.subsubsectionHead, .likesubsubsectionHead { font-size: 1.1487em; }
  @media (prefers-color-scheme: dark) { img[src^="speed-manual"]{filter: invert(1); } } 
 li p.indent { text-indent: 0em }
 li p:first-child{ margin-top:0em; }
@@ -168,6 +168,10 @@ div.author{white-space: nowrap;}
 div.abstract p {margin-left:5%; margin-right:5%;}
 div.abstract {width:100%;}
 .abstracttitle{text-align:center;margin-bottom:1em;}
+.subsectionToc, .likesubsectionToc {margin-left:1em;}
+.subsubsectionToc, .likesubsubsectionToc {margin-left:2em;}
+.paragraphToc, .likeparagraphToc {margin-left:3em;}
+.subparagraphToc, .likesubparagraphToc {margin-left:4em;}
 figure.float, div.figure {margin-left: auto; margin-right: auto;}
 figure.figure {text-align:center;}
 figcaption.caption {text-indent:-2em; margin-left:3em; margin-right:1em; text-align:center;}

src/README.md

@@ -35,6 +35,7 @@ run certcain things.
       + [Configuration and execution](#configuration-and-execution)
    * [CUDA](#cuda)
       + [Special Notes for sending CUDA jobs to the GPU Partition (`pg`)](#special-notes-for-sending-cuda-jobs-to-the-gpu-partition-pg)
+      + [Jupyter notebook example: Jupyter-Pytorch-CUDA](#jupyter-example-gpu-pytorch)
    * [Python Modules](#python-modules)
 
 <!-- TOC end -->
@@ -405,6 +406,21 @@ We have several versions of CUDA installed in:
 
 For CUDA to compile properly for the GPU queue, edit your `Makefile` replacing `/usr/local/cuda` with one of the above.
 
+<!-- TOC --><a name="jupyter-example-gpu-pytorch"></a>
+### Jupyter notebook example: Jupyter-Pytorch-CUDA
+
+Example prepared to run on speed, extracted from: https://developers.redhat.com/learning/learn:openshift-data-science:configure-jupyter-notebook-use-gpus-aiml-modeling/resource/resources:how-examine-gpu-resources-pytorch
+
+From speed-submit: 
+- Download `gpu-ml-model.ipynb` from this github to your `/speed-scratch/$USER space`
+- `salloc --mem=10Gb --gpus=1`
+
+From the node (interactive session):
+- `module load singularity/3.10.4/default`
+- `srun singularity exec -B $PWD\:/speed-pwd,/speed-scratch/$USER\:/my-speed-scratch,/nettemp --env SHELL=/bin/bash --nv /speed-scratch/nag-public/jupyter-pytorch-cuda.sif /bin/bash -c '/opt/conda/bin/jupyter notebook --no-browser --notebook-dir=/speed-pwd --ip="*" --port=8888 --allow-root'`
+- Follow the steps described in: https://nag-devops.github.io/speed-hpc/#jupyter-notebooks
+- When Jupyter is running on the browser, open `gpu-ml-model.ipynb` and run each cell
+
 <!-- TOC --><a name="python-modules"></a>
 ## Python Modules