Skip to content

[SYCL][ESIMD] Move headers to proper location.#3720

Merged
pvchupin merged 1 commit into
intel:syclfrom
kbobrovs:esimd_move_files
May 13, 2021
Merged

[SYCL][ESIMD] Move headers to proper location.#3720
pvchupin merged 1 commit into
intel:syclfrom
kbobrovs:esimd_move_files

Conversation

@kbobrovs
Copy link
Copy Markdown
Contributor

@kbobrovs kbobrovs commented May 11, 2021

Header location must reflect the contained API's namespace.
Not using CL as the top directory, as top-level cl namespace
has been deprecated in SYCL2020, so top-level CL include directory
will eventually be removed.

Signed-off-by: kbobrovs Konstantin.S.Bobrovsky@intel.com

@kbobrovs kbobrovs requested review from a team and DenisBakhvalov as code owners May 11, 2021 04:16
@kbobrovs kbobrovs requested a review from romanovvlad May 11, 2021 04:16
@kbobrovs kbobrovs force-pushed the esimd_move_files branch 2 times, most recently from 4ee8419 to 32c33ab Compare May 11, 2021 04:42
@kbobrovs
[SYCL][ESIMD] Move headers to proper location.
e625814 
Header location must reflect the contained API's namespace.
Not using CL as the top directory, as top-level cl namespace
has been deprecated in SYCL2020, so top-level CL include directory
will eventually be removed.

Signed-off-by: kbobrovs <Konstantin.S.Bobrovsky@intel.com>
@kbobrovs kbobrovs force-pushed the esimd_move_files branch from 32c33ab to e625814 Compare May 11, 2021 14:27
@kbobrovs
Copy link
Copy Markdown
Contributor Author

kbobrovs commented May 11, 2021 •
edited
Loading

The Jenkins failure is a known Gen vector BE problem (fixed in upcoming driver versions), so it should not prevent merging this PR.

[2021-05-11T16:36:18.232Z] $ "env" "SYCL_DEVICE_FILTER=opencl:gpu,host" ".../runDir/jenkins-dir/workspace/SYCL_CI/intel/Lin/LLVM_Test_Suite/build/SYCL/ESIMD/private_memory/Output/pm_access_1.cpp.tmp.out" "1"
[2021-05-11T16:36:18.232Z] # command output:
[2021-05-11T16:36:18.232Z] Running on Intel(R) HD Graphics 530 [0x1912]
[2021-05-11T16:36:18.232Z] 
[2021-05-11T16:36:18.232Z] # command stderr:
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTSN2cl4sycl6detail19__pf_kernel_wrapperI8KernelIDILi1EEEE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTS8KernelIDILi1EE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTSN2cl4sycl6detail19__pf_kernel_wrapperI8KernelIDILi2EEEE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTS8KernelIDILi2EE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTSN2cl4sycl6detail19__pf_kernel_wrapperI8KernelIDILi3EEEE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTS8KernelIDILi3EE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] LLVM ERROR: Cannot find pointer replacement

@kbobrovs kbobrovs requested a review from jbrodman May 11, 2021 20:03
DenisBakhvalov
DenisBakhvalov approved these changes May 11, 2021
View reviewed changes
Copy link
Copy Markdown
Contributor

@DenisBakhvalov DenisBakhvalov left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ESIMD part LGTM.

romanovvlad
romanovvlad reviewed May 12, 2021
View reviewed changes
Comment thread sycl/include/sycl/ext/intel/experimental/esimd.hpp
Comment on lines +17 to +18
#define SYCL_ESIMD_KERNEL
#define SYCL_ESIMD_FUNCTION
Copy link
Copy Markdown
Contributor

@romanovvlad romanovvlad May 12, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it possible to undef these macro at the end of the file?

Copy link
Copy Markdown
Contributor Author

@kbobrovs kbobrovs May 12, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The intent is actually make this available to user code - to mark ESIMD kernels and functions.

Copy link
Copy Markdown
Contributor

@romanovvlad romanovvlad May 12, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you please tell if this is documented then?

Copy link
Copy Markdown
Contributor Author

@kbobrovs kbobrovs May 12, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not, it is not. What is documented is the sycl_explicit_simd attribute, to which SYCL_ESIMD* macros are basically defined.

Copy link
Copy Markdown
Contributor Author

@kbobrovs kbobrovs May 12, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I acknowledge the problem wit documentation, but this will be solved in a separate PR. This one does not introduce these macros - it just moves from the deprecated header to the new one.

Comment thread sycl/include/sycl/ext/intel/experimental/esimd/detail/esimd_types.hpp
@kbobrovs kbobrovs requested a review from romanovvlad May 13, 2021 05:39
@bader
Copy link
Copy Markdown
Contributor

bader commented May 13, 2021

The Jenkins failure is a known Gen vector BE problem (fixed in upcoming driver versions), so it should not prevent merging this PR.

[2021-05-11T16:36:18.232Z] $ "env" "SYCL_DEVICE_FILTER=opencl:gpu,host" ".../runDir/jenkins-dir/workspace/SYCL_CI/intel/Lin/LLVM_Test_Suite/build/SYCL/ESIMD/private_memory/Output/pm_access_1.cpp.tmp.out" "1"
[2021-05-11T16:36:18.232Z] # command output:
[2021-05-11T16:36:18.232Z] Running on Intel(R) HD Graphics 530 [0x1912]
[2021-05-11T16:36:18.232Z] 
[2021-05-11T16:36:18.232Z] # command stderr:
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTSN2cl4sycl6detail19__pf_kernel_wrapperI8KernelIDILi1EEEE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTS8KernelIDILi1EE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTSN2cl4sycl6detail19__pf_kernel_wrapperI8KernelIDILi2EEEE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTS8KernelIDILi2EE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTSN2cl4sycl6detail19__pf_kernel_wrapperI8KernelIDILi3EEEE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] warning: GenXPromoteArray: _ZTS8KernelIDILi3EE allocation size is too big: using TPM
[2021-05-11T16:36:18.232Z] LLVM ERROR: Cannot find pointer replacement

Please, disable the failing test to avoid false alarms in other pull requests.
Is there any tracker for this issue already? If no, please, file one to re-enable the test once drivers are updated.

@kbobrovs
Copy link
Copy Markdown
Contributor Author

kbobrovs commented May 13, 2021

Please, disable the failing test to avoid false alarms in other pull requests.

The test is already disabled (recently): intel/llvm-test-suite@5d8eee5

Is there any tracker for this issue already? If no, please, file one to re-enable the test once drivers are updated.

Yes, there is internal tracker for this.

@pvchupin pvchupin merged commit 47aeafa into intel:sycl May 13, 2021
@kbobrovs kbobrovs deleted the esimd_move_files branch November 25, 2021 05:17
jsji pushed a commit that referenced this pull request May 14, 2026
@svenvh @sys-ce-bb
SPIRVReader: Add spirv.nontemporal for Nontemporal image op (#3720)
55e7ec0 
Add `spirv.nontemporal` metadata for image operations that have a
`Nontemporal` image operand, such that the nontemporal information is
preserved in the LLVM IR.

Original commit:
KhronosGroup/SPIRV-LLVM-Translator@8324c559351d1da
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@jbrodman jbrodman Awaiting requested review from jbrodman

2 more reviewers

@romanovvlad romanovvlad romanovvlad approved these changes

@DenisBakhvalov DenisBakhvalov DenisBakhvalov approved these changes

Reviewers whose approvals may not affect merge requirements

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

5 participants

@kbobrovs @bader @romanovvlad @DenisBakhvalov @pvchupin