Completed successfully sycl-web/sycl-latest-good -- dfd72fc - Fri Oct 10 19:35:17 2025 +0000 sycl-web/main-latest-good -- 24d5c4f - Sat Oct 11 04:40:33 2025 +0000

Merged to sycl-web with no conflict or build issue 

Build passed 

Merged to sycl-web with no conflict or build issue 

[SYCL][Doc] Simpler free function kernel enqueue (#19995) Specify a simplified way to enqueue free function kernels without first getting a kernel bundle. The style follows the enqueue functions in "sycl_ext_oneapi_enqueue_functions". Once the KHR extension "sycl_khr_free_function_commands" is finalized, we may also add enqueue functions that follow that style. However, we will first need to extend the `requirements` in that KHR to accept kernel launch properties.

[CI] Fail benchmark CI only on regressions in CPU instruction count (#… …20277) Benchmark script changes in this PR are NFC. Bit of a long overdue change: Wall time has proven to not be a good metric for stable performance numbers. This PR changes SYCL benchmarking CI regression filters to only fail on severe regressions in tests measuring CPU instruction count only. This PR does not affect the behavior of benchmarking scripts. It only affects the behavior of the CI w.r.t. regression failures for _SYCL_: it does not concern regressions detected in UR or L0.

[clang-linker-wrapper][NFC] Re-order command line arguments. (#20323) Set `llvm-link` parameters before input/output files. `llvm-link <params> <input files> -o <output file>`

[SYCL][E2E] Move free_function_kernels.cpp to UNSUPPORTED (#20305) It's causing all the hangs/fails on BMG in the [nightly](https://github.com/intel/llvm/actions/runs/18300912064/job/52109204455), I manually narrowed it down to this test. Nightly run with this change [here](https://github.com/intel/llvm/actions/runs/18324422627/job/52186913555). Signed-off-by: Sarnie, Nick <nick.sarnie@intel.com>

[UR][L0] Added safety checks to prevent double urProgramRelease (#20257) Users used to be able to call urProgramRelease on the same module, but with these changes, that would no longer be possible. --------- Signed-off-by: Zhang, Winston <winston.zhang@intel.com>

[SYCL][Graph] Reuse recording queue when available for finalize (#20106) There are two issues with the current implementation: - Creating a placeholder queue is adding overhead at `finalize()` to common path. - Queue properties are set to default when no queue can be reused at Graph `finalize()`.