llvm
llvm copied to clipboard
intel
Intel staging area for llvm.org contribution. Home for Intel LLVM-based projects.
Destructors are implicitly noexcept, so we must ensure they don't actually throw exceptions. No change to API or ABI with this PR.
This PR implements the traits and other APIs for free function support. The free function spec is here: https://github.com/intel/llvm/blob/sycl/sycl/doc/extensions/proposed/sycl_ext_oneapi_free_function_kernels.asciidoc. Later PRs will add the remaining parts of the spec. This...
When compiling for -fsycl-targets values of nvptx64-nvidia-cuda and amdgcn-amd-gpu, the default arch behaviors were not applied to the compilation. Updates to do the following: - Add default of sm_50 for...
### Describe the bug The double precision performance of [fft-sycl](https://github.com/zjin-lcf/HeCBench/tree/master/src/fft-sycl) lags behind [fft-cuda,](https://github.com/zjin-lcf/HeCBench/tree/master/src/fft-cuda) achieving only 70% of CUDA's performance. Profiling with nsys (`nsys nvprof --print-gpu-trace`) reveals that the SYCL code...
SYCL language support on the part of the kernel_compiler is specified here: https://github.com/intel/llvm/pull/11985 However, that specification is not presently realizable. We need more support from the FE and post link...
UR PR: https://github.com/oneapi-src/unified-runtime/pull/1806
Part of getting rid of the deprecated `exception::get_cl_code`.
Describe behavior of sycl-post-link in support of dynamic linking
