Use ZeroMemset in View constructor #4226
Merged
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
masterleinad
moved this from In progress
to Awaiting Feedback
in Developer: Daniel Arndt
DavidPoliakoff
suggested changes
Aug 11, 2021
| PolicyType policy(0, n); | ||
| std::string functor_name; | ||
| if (!space.in_parallel()) { |
There was a problem hiding this comment.
This is for use cases involving Views of Views, we need that branch, I believe
There was a problem hiding this comment.
But then it should still be a KOKKOS_FUNCTION, doesn't it? Anyway, I will just separate it into a different pull request.
There was a problem hiding this comment.
Sure. For reference, it isn't KOKKOS_FUNCTION, because the outer constructor still happens on the Host. It's really confusing, everything about View of Views is weird
DavidPoliakoff
approved these changes
Aug 11, 2021
crtrott
approved these changes
Aug 12, 2021
dalg24
reviewed
Aug 12, 2021
| @@ -783,7 +783,7 @@ KOKKOS_ADD_ADVANCED_TEST( UnitTest_PushFinalizeHook_terminate | |||
| EXE ProfilingAllCalls | |||
| TOOL kokkosprinter-tool | |||
| ARGS --kokkos-tools-args="-c test delimit" | |||
| PASS_REGULAR_EXPRESSION "kokkosp_init_library::kokkosp_parse_args:4:KokkosCore_ProfilingAllCalls:-c:test:delimit::.*::kokkosp_allocate_data:${MEMSPACE_REGEX}:source:${ADDRESS_REGEX}:40::kokkosp_begin_parallel_for:Kokkos::View::initialization [[]source]:0:0::kokkosp_end_parallel_for:0::kokkosp_allocate_data:${MEMSPACE_REGEX}:destination:${ADDRESS_REGEX}:40::kokkosp_begin_parallel_for:Kokkos::View::initialization [[]destination]:0:0::kokkosp_end_parallel_for:0::kokkosp_begin_deep_copy:${MEMSPACE_REGEX}:destination:${ADDRESS_REGEX}:${MEMSPACE_REGEX}:source:${ADDRESS_REGEX}:40::kokkosp_end_deep_copy::kokkosp_begin_parallel_for:parallel_for:${SIZE_REGEX}:0::kokkosp_end_parallel_for:0::kokkosp_begin_parallel_reduce:parallel_reduce:${SIZE_REGEX}:1${SKIP_SCRATCH_INITIALIZATION_REGEX}::kokkosp_end_parallel_reduce:1::kokkosp_begin_parallel_scan:parallel_scan:${SIZE_REGEX}:2::kokkosp_end_parallel_scan:2::kokkosp_push_profile_region:push_region::kokkosp_pop_profile_region::kokkosp_create_profile_section:created_section:3::kokkosp_start_profile_section:3::kokkosp_stop_profile_section:3::kokkosp_destroy_profile_section:3::kokkosp_profile_event:profiling_event::kokkosp_declare_metadata:dogs:good::kokkosp_deallocate_data:${MEMSPACE_REGEX}:destination:${ADDRESS_REGEX}:40::kokkosp_deallocate_data:${MEMSPACE_REGEX}:source:${ADDRESS_REGEX}:40::kokkosp_finalize_library::" | |||
| PASS_REGULAR_EXPRESSION "kokkosp_init_library::kokkosp_parse_args:4:KokkosCore_ProfilingAllCalls:-c:test:delimit::.*::kokkosp_allocate_data:${MEMSPACE_REGEX}:source:${ADDRESS_REGEX}:40::kokkosp_allocate_data:${MEMSPACE_REGEX}:destination:${ADDRESS_REGEX}:40::kokkosp_begin_deep_copy:${MEMSPACE_REGEX}:destination:${ADDRESS_REGEX}:${MEMSPACE_REGEX}:source:${ADDRESS_REGEX}:40::kokkosp_end_deep_copy::kokkosp_begin_parallel_for:parallel_for:${SIZE_REGEX}:0::kokkosp_end_parallel_for:0::kokkosp_begin_parallel_reduce:parallel_reduce:${SIZE_REGEX}:1.*::kokkosp_end_parallel_reduce:1::kokkosp_begin_parallel_scan:parallel_scan:${SIZE_REGEX}:2::kokkosp_end_parallel_scan:2::kokkosp_push_profile_region:push_region::kokkosp_pop_profile_region::kokkosp_create_profile_section:created_section:3::kokkosp_start_profile_section:3::kokkosp_stop_profile_section:3::kokkosp_destroy_profile_section:3::kokkosp_profile_event:profiling_event::kokkosp_declare_metadata:dogs:good::kokkosp_deallocate_data:${MEMSPACE_REGEX}:destination:${ADDRESS_REGEX}:40::kokkosp_deallocate_data:${MEMSPACE_REGEX}:source:${ADDRESS_REGEX}:40::kokkosp_finalize_library::" | |||
There was a problem hiding this comment.
@DavidPoliakoff @crtrott please confirm you looked at these changes carefully
There was a problem hiding this comment.
What happens here is that don't run a Kokkos::parallel_for anymore (but ZeroMemset) and thus also don't print related profiling information.
There was a problem hiding this comment.
This means we lose the ability to detect unnecessary view initialization with the tools
There was a problem hiding this comment.
Yes. Would you prefer us slightly lying by keeping the profiling information in place?
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.
Follow-up to #3944.
This pull request makes sure that we also use
std::memsetand similar in theViewconstruction.Since
ViewValueFunctor::executeis not aKOKKOS_FUNCTIONit doesn't make sense to ask forspace.in_parallel()IMHO, so I also removed that branch. I am happy to discuss that separately, of course.