Add documentation #8
Conversation
| on: | ||
| - push | ||
| # push: | ||
| # branches: | ||
| # - main |
There was a problem hiding this comment.
Needs to be fixed before merging.
| * [ETH Zurich]: Swiss Federal Institute of Technology Zurich | ||
| * [PASC]: Platform for Advanced Scientific Computing | ||
|
|
||
| <img height="50" src="./docs/images/logo-cscs.jpg"><img height="50" src="./docs/images/logo-eth.svg"><img height="50" src="./docs/images/logo-pasc.png"> |
There was a problem hiding this comment.
Require assets added in #6
|
Preliminary questions: |
There is a preview from my fork: https://rmeli.github.io/DLA-Future-Fortran
Not really... I'm open to suggestions. |
| project: DLA-Future-Fortran | ||
| summary: Fortan interface for DLA-Future | ||
| project_github: https://github.com/eth-cscs/DLA-Future-Fortran | ||
| author: Rocco Meli |
| ## Optimal parameters | ||
|
|
||
| ### Block size | ||
|
|
||
| | Device | Block size | | ||
| | :----: | :--------: | | ||
| | CPU | 512 | | ||
| | GPU | 1024 | | ||
|
|
||
| ### Ranks per node | ||
|
|
||
| For CPU-only calculations it is recommended to use one rank per NUMA domain. For GPU calculations it is recommended to use one rank per GPU. |
There was a problem hiding this comment.
I wonder if this is something we could put in DLA-Future's documentation instead and link to it from here? Additionally, I'd add some sort of disclaimer for the block sizes that these are typically the best sizes, but YMMV etc. etc. i.e do your own tests if you want "optimal parameters".
There was a problem hiding this comment.
Yes, adding this directly in DLA-Future's documentation would be ideal.
gh-pagesbranchDocumentation Preview
The documentation is currently incomplete: only one function per algorithm is documented, plus the helper functions. It will be completed once the phrasing is agreed upon.