[NFCI][SYCL] Use get_info_impl (not _nocheck) for USM info descriptors
#18453
Conversation
This file contains hidden or 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
…ptors I tracked the appearance of `_nocheck` down to intel#937 and James doesn't remember anything specific requiring that vs "checked" version. Also, some similar USM-related aspects (those that don't just delegate to `get_info`) do use "checked" APIs, so it seems logical to unify this processing. The main reason I do this is that it would be a bit easier to cache the values of "checked" interfaces by pre-computing USM support in `device_impl`'s ctor. Note that we perform at least some of those every time we allocate USM memory, so doing the caching is desirable (even for the sake of cleaning up our traces).
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
aelovikov-intel
changed the title
get_info_impl (not _nocheck) for USM info descri…get_info_impl (not _nocheck) for USM info descriptors
aelovikov-intel
commented
May 13, 2025
| @@ -589,33 +589,27 @@ class device_impl : public std::enable_shared_from_this<device_impl> { | |||
| } | |||
|
|
|||
| CASE(info::device::usm_device_allocations) { | |||
| return get_info_impl_nocheck<UR_DEVICE_INFO_USM_DEVICE_SUPPORT>() | |||
| .value_or(0) & | |||
| return get_info_impl<UR_DEVICE_INFO_USM_DEVICE_SUPPORT>() & | |||
There was a problem hiding this comment.
@intel/unified-runtime-reviewers , are any of these APIs expected to return a meaningful error that isn't an indication of everything being terribly broken?
There was a problem hiding this comment.
no, these aren't optional queries so it should be success or catastrophic failure as you say
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
aelovikov-intel
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
aarongreig
approved these changes
May 14, 2025
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.
I tracked the appearance of
_nocheckdown to#937 and James doesn't remember anything specific requiring that vs "checked" version. Also, some similar USM-related aspects (those that don't just delegate to
get_info) do use "checked" APIs, so it seems logical to unify this processing.The main reason I do this is that it would be a bit easier to cache the values of "checked" interfaces by pre-computing USM support in
device_impl's ctor. Note that we perform at least some of those every time we allocate USM memory, so doing the caching is desirable (even for the sake of cleaning up our traces).