rsnt_translations
56,430
edits
Advanced MPI scheduling: Difference between revisions
Marked this version for translation
No edit summary |
(Marked this version for translation) |
||
| Line 31: | Line 31: | ||
==== Whole nodes ==== <!--T:17--> | ==== Whole nodes ==== <!--T:17--> | ||
<!--T:23--> | |||
''Some 48-core nodes with Skylake processors are now available at [[Cedar]], and [[Niagara]] has 40-core nodes with hyperthreading. This section is awaiting revision to reflect these changes.'' | ''Some 48-core nodes with Skylake processors are now available at [[Cedar]], and [[Niagara]] has 40-core nodes with hyperthreading. This section is awaiting revision to reflect these changes.'' | ||
<!--T:24--> | |||
Most nodes in [[Cedar]] and [[Graham]] have 32 cores and 128GB or more of memory. If you have a large parallel job to run, which can use 32 or a multiple of 32 cores efficiently, you should request whole nodes like so: | Most nodes in [[Cedar]] and [[Graham]] have 32 cores and 128GB or more of memory. If you have a large parallel job to run, which can use 32 or a multiple of 32 cores efficiently, you should request whole nodes like so: | ||
--nodes=2 | --nodes=2 | ||