Make memory allocated by hipHostMalloc coarse-grained #5152
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Wrong target branch. |
JBludau
changed the title
|
cc @arghdos |
|
Retest this please. |
|
Failures (couple CUDA builds on machine that ran out of disk space) are unrelated and safe to ignore |
Co-authored-by: Nick Curtis arghdos
|
Retest this please |
dalg24
approved these changes
Jul 14, 2022
|
CI failure is unrelated (sycl time-out) |
Rombur
approved these changes
Jul 14, 2022
crtrott
approved these changes
Jul 14, 2022
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
hipHostMalloc allocates fine grained memory per default. According to the documentation of AMD there is no way of calling MemAdvise on it to make it coarse-grained (Nevertheless, the memAdvise works on some combinations of driver and hardware).
To get coarse-grained memory on all AMD GPUs we can allocate with the MemAttachGlobal flag (I will ask someone at AMD just to make sure).
The tests that verify if we get the memory grained-ness we want in the Kokkos::Views is part of the already merged PR 5112 as it was important for the HIPManagedSpace.