This repository was archived by the owner on Mar 28, 2023. It is now read-only.
[SYCL][CUDA] Add unit test for local arguments alignment #608
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
This issue was solved in intel/llvm#5113, local kernel arguments have to be aligned to the type size.
npmiller
requested review from
bader,
romanovvlad and
vladimirlaz
as code owners
|
/verify with intel/llvm#5113 |
vladimirlaz
approved these changes
Dec 12, 2021
vladimirlaz
approved these changes
Dec 14, 2021
steffenlarsen
approved these changes
Dec 21, 2021
bader
approved these changes
Jan 12, 2022
|
@npmiller it looks like the test is failing on Linux (OpenCL BE). I have restarted it with the latest compiler. |
vladimirlaz
suggested changes
Feb 11, 2022
There was a problem hiding this comment.
@npmiller could you please fix CI issue reported for OpenCL BE.
[2022-02-10T07:00:45.849Z] ******************** TEST 'SYCL :: Regression/local-arg-align.cpp' FAILED ********************
[2022-02-10T07:00:45.849Z] Script:
[2022-02-10T07:00:45.849Z] --
[2022-02-10T07:00:45.849Z] : 'RUN: at line 1'; /runDir/jenkins-dir/workspace/llvm.obj/bin/clang++ -fsycl -fsycl-targets=spir64 /runDir/jenkins-dir/workspace/llvm-test-suite/SYCL/Regression/local-arg-align.cpp -o /runDir/jenkins-dir/workspace/build/SYCL/Regression/Output/local-arg-align.cpp.tmp.out
[2022-02-10T07:00:45.849Z] : 'RUN: at line 3'; env SYCL_DEVICE_FILTER=host /runDir/jenkins-dir/workspace/build/SYCL/Regression/Output/local-arg-align.cpp.tmp.out
[2022-02-10T07:00:45.849Z] : 'RUN: at line 4'; env SYCL_DEVICE_FILTER=opencl:cpu,host /runDir/jenkins-dir/workspace/build/SYCL/Regression/Output/local-arg-align.cpp.tmp.out
[2022-02-10T07:00:45.849Z] : 'RUN: at line 5'; env SYCL_DEVICE_FILTER=opencl:gpu,host /runDir/jenkins-dir/workspace/build/SYCL/Regression/Output/local-arg-align.cpp.tmp.out
[2022-02-10T07:00:45.849Z] : 'RUN: at line 6'; env SYCL_DEVICE_FILTER=acc /runDir/jenkins-dir/workspace/build/SYCL/Regression/Output/local-arg-align.cpp.tmp.out
[2022-02-10T07:00:45.849Z] --
[2022-02-10T07:00:45.849Z] Exit Code: 255
[2022-02-10T07:00:45.849Z]
[2022-02-10T07:00:45.849Z] Command Output (stdout):
[2022-02-10T07:00:45.849Z] --
[2022-02-10T07:00:45.849Z] $ ":" "RUN: at line 1"
[2022-02-10T07:00:45.849Z] note: command had no output on stdout or stderr
[2022-02-10T07:00:45.849Z] $ "/runDir/jenkins-dir/workspace/llvm.obj/bin/clang++" "-fsycl" "-fsycl-targets=spir64" "/runDir/jenkins-dir/workspace/llvm-test-suite/SYCL/Regression/local-arg-align.cpp" "-o" "/runDir/jenkins-dir/workspace/build/SYCL/Regression/Output/local-arg-align.cpp.tmp.out"
[2022-02-10T07:00:45.849Z] note: command had no output on stdout or stderr
[2022-02-10T07:00:45.849Z] $ ":" "RUN: at line 3"
[2022-02-10T07:00:45.849Z] note: command had no output on stdout or stderr
[2022-02-10T07:00:45.849Z] $ "env" "SYCL_DEVICE_FILTER=host" "/runDir/jenkins-dir/workspace/build/SYCL/Regression/Output/local-arg-align.cpp.tmp.out"
[2022-02-10T07:00:45.849Z] # command output:
[2022-02-10T07:00:45.849Z] Error: incorrect alignment for argument b, required alignment: 32, address: 0x1af3730
[2022-02-10T07:00:45.849Z]
[2022-02-10T07:00:45.849Z] error: command failed with exit status: 255
[2022-02-10T07:00:45.849Z]
[2022-02-10T07:00:45.849Z] --
[2022-02-10T07:00:45.849Z]
[2022-02-10T07:00:45.849Z] ********************
|
Oh that's interesting, do you have more details on what OpenCL runtime and device is being used in the CI? I've had another look through the test, and I can't really see anything wrong with it, I think this is likely a problem in the OpenCL runtime, or potentially how the SPIR-V binaries are generated, I'll try to see if I can track it down a bit further. |
|
I'm actually able to reproduce this with the Nvidia OpenCL so I reckon this is likely what the CI also uses. |
|
Okay, so I have tracked this down, it's actually not an issue with OpenCL at all, it's a host problem. Local accessors for host are allocated as Which is exactly what happens in this test. Replacing this with an aligned allocation on |
bader
pushed a commit
to intel/llvm
that referenced
this pull request
Mar 1, 2022
Local kernel arguments must be aligned to the type size, simply using `std::vector<char>` doesn't always provide the correct alignment. So this patch adds extra padding to the vector and ensures that the pointer returned for the accessor is actually aligned to the type size. This issue was exposed by: intel/llvm-test-suite#608, which was a follow up to fixing local accessor alignment for the CUDA plugin.
|
The failing test should be fixed now that intel/llvm#5554 was merged |
vladimirlaz
approved these changes
Mar 1, 2022
|
will wait till next compiler nightly to ensure that the test will pass with it before merge |
steffenlarsen
approved these changes
Mar 1, 2022
myler
added a commit
to myler/llvm-test-suite
that referenced
this pull request
Apr 12, 2022
CMPLRTST-15545: fixed git related issue, test driver is able to identify good commit.
myler
pushed a commit
to myler/llvm-test-suite
that referenced
this pull request
Apr 12, 2022
This issue was solved in intel/llvm#5113, local kernel arguments have to be aligned to the type size.
aelovikov-intel
pushed a commit
to aelovikov-intel/llvm
that referenced
this pull request
Mar 27, 2023
…test-suite#608) This issue was solved in intel#5113, local kernel arguments have to be aligned to the type size.
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
This issue was solved in intel/llvm#5113, local kernel arguments have to
be aligned to the type size.
cc @romanovvlad