DPC++ daily 2022-06-10

[SYCL][CUDA] Add experimental context and device interop (#6202)

This PR is adds part of the CUDA-backend spec interop proposed in KhronosGroup/SYCL-Docs#197. The changes work with the CUDA CTS interop checks KhronosGroup/SYCL-CTS#336.

This PR just adds the context and device interop. Further PRs will be submitted to fill in other missing interop functionality.

The changes are experimental so as to not break oneMKL, and oneDNN which use CUDA interops.

The experimental interop is enabled by defining the SYCL_EXT_ONEAPI_BACKEND_CUDA_EXPERIMENTAL macro and including the header file sycl/ext/oneapi/experimental/backend/cuda.hpp.

llvm-test-suite: intel/llvm-test-suite#1041

DPC++ daily 2022-06-09

sycl-nightly/20220609

[SYCL] Fix unused value warning (#6276)

DPC++ daily 2022-06-08

[SYCL] Isolate OpenCL types and enums (#6261)

This PR populates PI enums and typedefs with values taken from the OpenCL equivalents; it also makes sure that all uses of the system OpenCL header `cl.h` are channelled through `include/CL/sycl/detail/cl.h`. Finally, `pi.h` and `pi.hpp` would only tap into the system `cl.h` for OpenCL backend (guarded by `PI_OPENCL_AVAILABLE` compiler definition), which allowed a removal of all OpenCL types and enums from other PI backends.

Fixes: https://github.com/intel/llvm/issues/1265

DPC++ daily 2022-06-07

[GHA] Uplift GPU RT version for Nightly Builds (#6260)

Co-authored-by: GitHub Actions <[email protected]>

DPC++ daily 2022-06-06

sycl-nightly/20220606

[SYCL] Change f[no-]sycl-device-lib-online-link to f[no-]sycl-device-…

DPC++ daily 2022-06-05

LLVM and SPIRV-LLVM-Translator pulldown (WW24)

LLVM: https://github.com/llvm/llvm-project/commit/dd6bcdbf21716c56d3defd7f4cacddc7befd5de1
SPIRV-LLVM-Translator: https://github.com/KhronosGroup/SPIRV-LLVM-Translator/commit/24c5671

DPC++ daily 2022-06-04

[SYCL] Align usm_allocator with the specification (#6191)

According to

https://www.khronos.org/registry/SYCL/specs/sycl-2020/html/sycl-2020.html#_c_allocator_interface

the default value for the "Alignment" template parameter must be 0. Otherwise
the rebound allocator wouldn't change the alignment properly.

Test added in https://github.com/intel/llvm-test-suite/pull/1038

DPC++ daily 2022-06-03

[SYCL][HIP] Use valid cast for CUdeviceptr in HIP PI (#6207)

Replace invalid reinterpret_cast with static_cast
for CUdeviceptr in hip_piextMemGetNativeHandle
when building HIP PI on CUDA machine;

Signed-off-by: Lukas Sommer <[email protected]>

DPC++ daily 2022-06-02

sycl-nightly/20220602

[SYCL] fflush(stderr) to deliver assert message before abort'ing (#6227)

DPC++ daily 2022-06-01

sycl-nightly/20220601

[SYCL][L0] Force submit of dependent open command batches (#6199)