You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@tjdasso has done some great work tackling #1324. As soon as core/mem/disk specification is merged (the core specification is already implemented in his branch), we will have the ability to tell WQ specific resource requirements for a task. A next step that I think would be extremely useful would be to implement the missing pieces in order to make WQ aware of what category the task is in (as a first pass, we can try just using the app name as a category). This will allow us to take advantage of WQ's ability to optimally size tasks, using the specified resource requirements as a first guess.
For another project I'm implementing some bioinformatics pipelines, so I would be an enthusiastic tester for this use case.
The text was updated successfully, but these errors were encountered:
I have a PR in (#1677) that does exactly what @annawoodard proposes. I added two options: one to automatically assign a fresh category for each function, and one to turn on WQ's automatic resource labeling. With both of these turned on, WQ should be able to do a pretty OK job of figuring out resource requirements automatically and subdividing large workers.
@tjdasso has done some great work tackling #1324. As soon as core/mem/disk specification is merged (the core specification is already implemented in his branch), we will have the ability to tell WQ specific resource requirements for a task. A next step that I think would be extremely useful would be to implement the missing pieces in order to make WQ aware of what category the task is in (as a first pass, we can try just using the app name as a category). This will allow us to take advantage of WQ's ability to optimally size tasks, using the specified resource requirements as a first guess.
For another project I'm implementing some bioinformatics pipelines, so I would be an enthusiastic tester for this use case.
The text was updated successfully, but these errors were encountered: