[SYCL] Add force range rounding option and introduce new compiler flag #12715
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
ldrumm
requested changes
Feb 14, 2024
hdelan
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
mdtoguchi
reviewed
Feb 15, 2024
hdelan
had a problem deploying
to
WindowsCILock
GitHub Actions
Failure
premanandrao
approved these changes
Feb 15, 2024
mdtoguchi
reviewed
Feb 16, 2024
mdtoguchi
reviewed
Feb 16, 2024
|
CUDA should fail because of additional checks in |
bso-intel
approved these changes
Mar 11, 2024
This commit adds a new preference for range rounding, force, such that if the compile flag is used, only the range rounded parallel_for kernel will be generated. This can make binaries smaller as there is no duplication of SYCL range kernels across range rounded and unrounded versions. I have also added the flag: -fsycl-range-rounding, which can have values: on, force or disable. This flag aims to supercede the fsycl-disable-range-rounding flag. I have also added to existing tests to check for the functionality of the new flag and refactored the range rounding sycl-e2e test.
Add some description of how the -fsycl-range-rounding flag should be used.
- Change if else to switch in integration header emission and init preprocessor - Change comment in handler.hpp - Change comments and use static_assert with message in test-e2e - Change enum to have no defined int values - Wrap long line in LangOptions.def #
Makes -fsycl-range-rounding= accessible to driver calls, not just cc1 invocations.
Range rounding disable is tested in another test.
Range rounding is disabled for -O0 but now a user preference for range rounding can override this.
Make sure that if -fsycl-range-rounding=force is used, there is no emission of the unrounded range kernel at -O0 and -Od
hdelan
force-pushed
the
range-rounding-preferences
branch
from
d77cc74 to
d0b0d75
Compare
|
Ping @mdtoguchi for review, thanks |
|
Ping @intel/llvm-gatekeepers this can be merged |
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.
Adds a new preference for range rounding, force, such that if the compile flag is used, only the range rounded parallel_for kernel will be generated. This can make binaries smaller as there is no duplication of SYCL range kernels across range rounded and unrounded versions.
I have also added the flag: -fsycl-range-rounding, which can have values: on, force or disable. This flag aims to supercede the fsycl-disable-range-rounding flag.
I have also added to existing tests to check for the functionality of the new flag and refactored the range rounding sycl-e2e test. Also added brief description of flag's behaviour in
doc