pulldown from Sycl on 20250704 #2929
Open
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
…#18979) Part of the ongoing refactoring to prefer raw ptr/ref for SYCL RT objects by default with explicit `shared_from_this` when lifetimes need to be extended.
These should have the potential to be more optimal for most targets, as they emit vectorised codegen. These were the last functions that tripped the bitwise conditional parenthesis warning in libspirv, so that has been re-enabled.
Enabling these (or any extensions, really) by default is incorrect and prevents the r600 OpenCL builtins from being built. Each clang target reports which OpenCL extensions it does or doesn't support. This commit also moves the flag enabling SPIR-V builtin declarations to the SPIR-V options. They are not needed for the compilation of CLC or OpenCL builtins. There is no change to any builtins file.
The `-fsycl-device-lib-jit-link` only has an effect in SPIR compilation so these tests would fail with `-Werror`. It might be worth re-working these tests a bit to only run this part on spir, but other existing tests already have this change to make it work on `-Werror` so do it as well here to fix the nightly. Fixes intel/llvm#18744
`sycl_ext_oneapi_graph` is not yet supported on Windows for PTL devices, see CMPLRTST-27275
Add a basic implementation of buffers to the Offload adapter. The design is loosely based on the CUDA and HIP adapters. For now contexts only support a single device, so the implementation is relatively simple. When multi-device contexts are supported we will need to handle migration of data between devices (or otherwise change this implementation)
…ssor` from required testing items outlined in the test plan for free function kernels extension (#18994) This PR removes `sampled_image_accessor` and `unsampled_image_accessor` from required testing items outlined in the test plan for free function kernels extension. It's is done because `sampled_image_accessor` and `unsampled_image_accessor` are not supported by any device and they're only supported by host.
Our device compiler have been capable of automagically declaring necessary SPIR-V built-ins on the fly for a while now, meaning that we don't need them to be forward-declared in headers. This PR drops includes of `__spirv_ops.hpp` so that they don't appear anymore in `core.hpp` (and some other headers). The header is not removed entirely, however, because not every built-in is known to the compiler, i.e. some of them still have to be forward-declared in the header. Most likely there are other places which can be made free of uses of the header and the header itself can probably be cleaned up agressively, but I will leave it for separate future PRs. --------- Co-authored-by: Udit Kumar Agarwal <[email protected]>
This commit adds a new property to the property set when producing SYCLBIN files. This new property lists all kernels inside the corresponding binaries. --------- Signed-off-by: Larsen, Steffen <[email protected]> Co-authored-by: Marcos Maronas <[email protected]>
This is required to ensure that the correct target with the correct options is used when linking with clang in the test. For example without trying to use the `SPIR-V` (not the SPIR) backend for testing with `--param spirv-backend=True` does not take effect. If the test would be run on a non SPIR target it would fail as clang is expecting object files for the SPIR target when `-fsycl-targets` is not set.
…983) Part of the ongoing refactoring to prefer raw ptr/ref for SYCL RT objects by default with explicit `shared_from_this` when lifetimes need to be extended.
The NVPTX target was unable to properly optimize the global ID query, despite the user specifying the -fsycl-id-queries-fit-in-int flag. This is because, once linked, the compiler sees the global ID builtin as (i64 add (mul (i64 zext i32 A), (i64 zext i32 B), (i64 zext i32 C))). Despite knowing that each of A, B and C are 32-bit values, and the final result fits in a 32-bit value, it is not legal to replace this sequence with (i64 zext (add i32 (mul i32 A, B), C)), which is the ideal code here. The solution to this problem is a new opt-in 'reflection' in the NVPTX implementation of the global ID builtin, which selects a more optimal version. The driver enables this reflection only when the user passes -fsycl-id-queries-fit-in-int.
…ether (#18992) The correct test is already there: https://github.com/intel/llvm/blob/253228255b8c56bc5d4d8919a2020bcc7847f1ac/clang/test/Driver/fsyclbin.cpp#L8-L14
This enables building rand for CUDA and HIP and enables the test.
Continuation of the refactoring efforts in intel/llvm#18715 intel/llvm#18748 intel/llvm#18830 intel/llvm#18907 intel/llvm#18983
This test is passing now.
This PR 1. adds **get_kernel_info** functions for kernel free functions, [docs](https://github.com/intel/llvm/blob/sycl/sycl/doc/extensions/proposed/sycl_ext_oneapi_free_function_kernels.asciidoc#new-free-functions-to-query-kernel-information-descriptors) 2. fixes bug when more than one property added with **add_ir_attributes_funcion**, free function was not recognized as a kernel function
The queue implementation assumed that handler.finalize() will return either a valid (non-discarded) event or nullptr (that's why parseEvent is nop in preview mode). However, that was not always true. It was possible for scheduler to only mark an event as discarded after handler.finalize() completed (during actual command execution). This resulted in discarded event being stored in LastEventPtr in the queue and calls to wait() on that discarded event (which is not allowed) in MT scenarios. Fix this, by modyfing addCG() to mark the event as discarded immediately and to return nullptr is the event is discarded. --------- Co-authored-by: aelovikov-intel <[email protected]>
When releasing executionEvent in the commandBuffer we need to also retain the queue. Otherwise event->release() in commandBuffer's destructor might attempt to release the event to an event pool that has already been destroyed. Also, make sure eventPool is destroyed after commandListManager to avoid any issues with events owned by the commandListManager and move context retain/release from commandListManager to queue/command_buffer. The implementation of commandListManager marked move ctor and assignment operator as defaulted which was wrong - the implementation would have to account for context and device retain/release calls. To avoid this, move the logic to queue/commandBuffer which can have move ctors/assignments operators removed.
- Remove switch case for 8,16,32 now that all format int sizes are supported. Signed-off-by: Neil R. Spruit <[email protected]>
…026) Bumps [brace-expansion](https://github.com/juliangruber/brace-expansion) from 1.1.11 to 1.1.12. <details> <summary>Release notes</summary> <p><em>Sourced from <a href="https://github.com/juliangruber/brace-expansion/releases">brace-expansion's releases</a>.</em></p> <blockquote> <h2>v1.1.12</h2> <ul> <li>pkg: publish on tag 1.x c460dbd</li> <li>fmt ccb8ac6</li> <li>Fix potential ReDoS Vulnerability or Inefficient Regular Expression (<a href="https://redirect.github.com/juliangruber/brace-expansion/issues/65">#65</a>) c3c73c8</li> </ul> <hr /> <p><a href="https://github.com/juliangruber/brace-expansion/compare/v1.1.11...v1.1.12">https://github.com/juliangruber/brace-expansion/compare/v1.1.11...v1.1.12</a></p> </blockquote> </details> <details> <summary>Commits</summary> <ul> <li><a href="https://github.com/juliangruber/brace-expansion/commit/44f33b47c5c6a965d507421af43e86cf5971d711"><code>44f33b4</code></a> 1.1.12</li> <li><a href="https://github.com/juliangruber/brace-expansion/commit/c460dbd68e428d147b2080622d8ce126c7a08570"><code>c460dbd</code></a> pkg: publish on tag 1.x</li> <li><a href="https://github.com/juliangruber/brace-expansion/commit/ccb8ac6d4292b7661b677fe048ba6690c877f51f"><code>ccb8ac6</code></a> fmt</li> <li><a href="https://github.com/juliangruber/brace-expansion/commit/c3c73c8b088defc70851843be88ccc3af08e7217"><code>c3c73c8</code></a> Fix potential ReDoS Vulnerability or Inefficient Regular Expression (<a href="https://redirect.github.com/juliangruber/brace-expansion/issues/65">#65</a>)</li> <li>See full diff in <a href="https://github.com/juliangruber/brace-expansion/compare/1.1.11...v1.1.12">compare view</a></li> </ul> </details> <br /> [](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores) Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting `@dependabot rebase`. [//]: # (dependabot-automerge-start) [//]: # (dependabot-automerge-end) --- <details> <summary>Dependabot commands and options</summary> <br /> You can trigger Dependabot actions by commenting on this PR: - `@dependabot rebase` will rebase this PR - `@dependabot recreate` will recreate this PR, overwriting any edits that have been made to it - `@dependabot merge` will merge this PR after your CI passes on it - `@dependabot squash and merge` will squash and merge this PR after your CI passes on it - `@dependabot cancel merge` will cancel a previously requested merge and block automerging - `@dependabot reopen` will reopen this PR if it is closed - `@dependabot close` will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually - `@dependabot show <dependency name> ignore conditions` will show all of the ignore conditions of the specified dependency - `@dependabot ignore this major version` will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this minor version` will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself) - `@dependabot ignore this dependency` will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself) You can disable automated security fix PRs for this repo from the [Security Alerts page](https://github.com/intel/llvm/network/alerts). </details> Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
…ration header (#18929) This PR fixes bug with templated kernel free function and its specialization, i.e. forward declaration of specialization is emitted into integration header too.
This feature is not supported on CUDA and HIP. There was only one test without the unsupported directive, the `free_function_kernels.cpp` test, but this is also tested in `KernelAndProgram` without the image dependency flag and more thoroughly. So since the feature is unsupported anyway it makes sense to disable it as well.
…ress space (#19011) In LowerWGScope pass, getSizeTTy is used as return type of local id functions. We should use pointer size in global address space, to ensure size_t is sufficient to represent the global range.
This PR allows changes in intel/llvm#19280 to be tested. This PR will be open for merging if changes in #19280 are approved.
To fix CTS test (https://github.com/KhronosGroup/SYCL-CTS/blob/b5cf985ecc497f82045165e9309fdc7a990c7e60/tests/extension/khr_queue_empty_query/queue_empty.cpp) failures for the `sycl_khr_queue_empty_query` extension (implemented in intel/llvm#18799), which are caused by the error `UR_RESULT_ERROR_UNSUPPORTED_ENUMERATION` emitted in the old code. --------- Signed-off-by: Hu, Peisen <[email protected]>
Forgot to add manual dispatch in #19281
Add intel_cpu_spr There is a test already, but we don't launch SPR&CPU. Tested locally.
When NumDevices is not zero and when phDevice is null, the user cannot query the number of devices. This addresses that issue. --------- Signed-off-by: Zhang, Winston <[email protected]>
Auto merge from sycl. Conflicts to be resolved.
Fix conflicts
zhiweij1
reviewed
Jul 7, 2025
zhiweij1
reviewed
Jul 7, 2025
zhiweij1
reviewed
Jul 7, 2025
zhiweij1
reviewed
Jul 7, 2025
zhiweij1
reviewed
Jul 7, 2025
zhiweij1
reviewed
Jul 7, 2025
tomflinda
reviewed
Jul 7, 2025
Address the review comments
zhiweij1
approved these changes
Jul 7, 2025
tomflinda
approved these changes
Jul 8, 2025
There was a problem hiding this comment.
Local verification passed(including build pass+ lit tests pass)
[83/84] Running lit suite /mnt/hdd/Chenwei/Repo/SYCLomatic/clang/test/dpct
llvm-lit: /mnt/hdd/Chenwei/Repo/SYCLomatic/llvm/utils/lit/lit/llvm/config.py:520: note: using clang: /mnt/hdd/Chenwei/Repo/build_release/bin/clang
llvm-lit: /mnt/hdd/Chenwei/Repo/SYCLomatic/llvm/utils/lit/lit/llvm/subst.py:126: note: Did not find cir-opt in /mnt/hdd/Chenwei/Repo/build_release/bin:/mnt/hdd/Chenwei/Repo/build_release/bin
Testing Time: 1158.84s
Total Discovered Tests: 1277
Unsupported: 73 (5.72%)
Passed : 1204 (94.28%)
ziranzha
approved these changes
Jul 8, 2025
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.
pulldown from Sycl on 20250704.