cc_staff
653
edits
Talk:Running jobs: Difference between revisions
thoughts on memory management
No edit summary |
(thoughts on memory management) |
||
| Line 2: | Line 2: | ||
The requirement for <code>--account=project-name-cpu</code> is still under discussion and may change before release. | The requirement for <code>--account=project-name-cpu</code> is still under discussion and may change before release. | ||
:[[User:Rdickson|Ross Dickson]] ([[User talk:Rdickson|talk]]) 13:55, 3 April 2017 (UTC) | :[[User:Rdickson|Ross Dickson]] ([[User talk:Rdickson|talk]]) 13:55, 3 April 2017 (UTC) | ||
==Memory Management== | |||
* According to Kamil it is advisable for users to express their job in 1000s on MB instead of GB (i.e. 1000MB ~ 1GB). This will leave some RAM for the OS when reaching the memory limit on core nodes. e.g. <code>--ntasks=32 --mem-per-cpu=4000M</code> will fit on a base node with 128G of RAM, while <code>--ntasks=32 --mem-per-cpu=4G</code> requires a Large-memory node. | |||
* At least on Graham, one needs to use <code>--mem=xxxG</code> to request a large-memory node and with using <code>--mem-per-cpu=yyyG</code> one can only ever get base-nodes. | |||
:[[User:Stuekero|Oliver Stueker]] ([[User talk:Stuekero|talk]]) 19:25, 11 July 2017 (UTC) | |||
==External links== | ==External links== | ||