-
Notifications
You must be signed in to change notification settings - Fork 407
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Remove Experimental::LogicalMemorySpace
#6557
Conversation
@PhilMiller voiced an objection to this in #6555 (comment)
|
Thanks for copying that over - I was on my phone and hadn't seen this PR put up yet. Magistrate is here: https://github.com/DARMA-tasking/magistrate/ I realize it's disabled at the moment, but as I understand (having stepped away from the project) there are broader design changes being done in the stuff that it would be testing. It should still be applicable. One of the longer-standing NGA folks should be able to follow up on that point. |
Pinging @ElisabethGiem as this might be of interest for her due to her work with resilient execution spaces |
@cz4rs since you worked a bit on magistrate, any comments on this? |
I would prefer to just deprecate. After all, the class is pretty isolated and it's not obvious that its existence would impact any refactoring or cause any other problems. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We agreed in the team meeting to remove LogicalMemorySpace
without deprecating it. The removal looks fine to me.
since Kokkos::Experimental::LogicalMemorySpace was removed in kokkos/kokkos#6557
…s::Experimental::LogicalMemorySpace was removed in kokkos/kokkos#6557
…s::Experimental::LogicalMemorySpace was removed in kokkos/kokkos#6557
…s::Experimental::LogicalMemorySpace was removed in kokkos/kokkos#6557
…s::Experimental::LogicalMemorySpace was removed in kokkos/kokkos#6557
Proposed resolution for #5609
In short, coming to the conclusion that
LogicalMemorySpace
cannot achieve its design goal (that is a debugging facility for accessibility issues on systems without a discrete GPU) and suggesting to just remove it.The code sits in the
Experimental::
namespace. I am not aware of any serious use of this facility.I did a quick search on GH and AFAICT the only code we'd break is kokkos-remote-spaces but we could fix it ourselves. It looks like Kokkos-resillience would not be affected by that change. @janciesko @nmm0
Let me know if you think I am too aggressive and that we should consider going through deprecation.