cc_staff
812
edits
Allocations and compute scheduling/fr: Difference between revisions
Jump to navigation
Jump to search
Allocations and compute scheduling/fr (view source)
Revision as of 14:47, 5 September 2023
, 1 year agoCreated page with "=Unités GPU de référence= {{Note|Cette nouvelle unité sera utilisée à partir du CAR 2024.}}"
No edit summary Tags: Mobile edit Mobile web edit |
(Created page with "=Unités GPU de référence= {{Note|Cette nouvelle unité sera utilisée à partir du CAR 2024.}}") Tags: Mobile edit Mobile web edit |
||
| Line 28: | Line 28: | ||
Il se pourrait qu’au cours d’un mois ou d’une année vous puissiez accomplir plus de travail que votre allocation ne semblerait le permettre, mais ce scénario est peu probable puisque la demande est plus élevée que la quantité de ressources dont nous disposons. | Il se pourrait qu’au cours d’un mois ou d’une année vous puissiez accomplir plus de travail que votre allocation ne semblerait le permettre, mais ce scénario est peu probable puisque la demande est plus élevée que la quantité de ressources dont nous disposons. | ||
= | =Unités GPU de référence= | ||
{{Note| | {{Note|Cette nouvelle unité sera utilisée à partir du CAR 2024.}} | ||
As you may be aware, the performance of GPUs has dramatically increased in the recent years and is expected to do so again with the upcoming next generation of GPUs. Until RAC 2023, in order to reduce complexity, we have been treating all GPUs as equivalent to each other at allocation time and when considering how many resources groups have consumed. This has raised issues of fairness, both in the allocation process and while running jobs. We cannot continue to treat all GPU types as the same. | As you may be aware, the performance of GPUs has dramatically increased in the recent years and is expected to do so again with the upcoming next generation of GPUs. Until RAC 2023, in order to reduce complexity, we have been treating all GPUs as equivalent to each other at allocation time and when considering how many resources groups have consumed. This has raised issues of fairness, both in the allocation process and while running jobs. We cannot continue to treat all GPU types as the same. | ||