[SYCL-MLIR] pulldown (20230531) #9666
Merged
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
clang-offload-bundler functionality was taken out from tool into "library" here: https://reviews.llvm.org/D129873. This commit adjusts CODEOWNERS file so that DPC++ tools team could track changes.
Signed-off-by: Michael Aziz <michael.aziz@intel.com>
The existing implementation of host-side abs_diff relies on undefined behavior when the signed operations overflow. This can mean that using different compilers to build DPC++ may break the implementation. This commit changes the implementation to avoid this undefined behavior. --------- Signed-off-by: Larsen, Steffen <steffen.larsen@intel.com> Co-authored-by: Dmitry Vodopyanov <dmitry.vodopyanov@intel.com>
Signed-off-by: Michael Aziz <michael.aziz@intel.com>
…l#9572) AOCO specific objects can be added to an archive. When this occurs, the archive was being completely ignored, causing issues with device compilation. Take advantage of the additional `-excluded-targets` option from the `clang-offload-bundler` so we can process these archives for the needed device information, excluding potential unbundling of the target device in the presence of `aoco` binaries in that same object. --------- Co-authored-by: Michael Toguchi <michael.d.toguchi@intel.com>
…tel#9612) This property is queried by sycl-ls tool.
This small patch fixes the TK size for tf32 type.
…intel#9624) Using (cu|hip)DeviceGetPCIBusId API call, present since at least CUDA 8.0 / ROCm 2.8. Unlike L0, we do not require setting `SYCL_ENABLE_PCI=1` environment variable.
…9606) sycl_devices is used instead now, including our public and internal CI.
Adds prefetch and joint_prefetch functions for providing prefetch hints associated with specific levels of cache. --------- Signed-off-by: John Pennycook <john.pennycook@intel.com>
Signed-off-by: Jaime Arteaga <jaime.a.arteaga.molina@intel.com>
…ut sub groups (intel#9543) Change the fallback behavior for `PI_KERNEL_MAX_SUB_GROUP_SIZE` to return the maximum sub group size for the device, rather than the maximum work group size for the kernel, on devices where sub groups are not supported. This change makes it consistent with the result of the `PI_DEVICE_INFO_SUB_GROUP_SIZES_INTEL` query. --------- Signed-off-by: Michael Aziz <michael.aziz@intel.com>
This patch adds a new interface that can be used to lookup the required accuracy of floating-point builtin intrinsics for a specified type and accuracy level (high, medium, low, sycl, or cuda). A later change will add support for this functionality in the IRBuilder class.
I believe the issue has been introduced in intel#9572.
pi_bool is uint32_t and ur_bool_t is uint8_t, so to make sure correct functionality is maintained, use uint32_t as replacement for pi_bool, instead of ur_bool_t. Signed-off-by: Jaime Arteaga <jaime.a.arteaga.molina@intel.com>
And use that as dynamic features in our SYCL End-to-End tests framework.
…ly` accessor (intel#9651) * Modifies accessor::ConcreteASPtrType to be `const` for `readonly` accessor to fix failing case where copying accessor<const T>::MData does not compile when src type is `const` and dest is `non-const`. * Add test
…#9642) check_has.cpp strictly depends on order of sycl_declared_aspects and srcloc metadata which is not guaranteed by compiler. This test aims to check IR for device_has(aspect..., we can keep sycl_declared_aspects and remove srcloc to avoid the dependency. --------- Signed-off-by: jinge90 <ge.jin@intel.com>
sys-ce-bb
added
disable-lint
… `readonly` accessor (intel#9651)" This reverts commit fd45980.
whitneywhtsang
approved these changes
May 31, 2023
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.
Merged from sycl.
Reverted fd45980 because it depends on 712cb4e.