llvm
llvm copied to clipboard
intel
Intel staging area for llvm.org contribution. Home for Intel LLVM-based projects.
On a CUDA & Windows setup when shared USM is used there is an issue with using the allocations concurrently in both device commands and host-tasks. This is based on...
https://github.com/intel/llvm/pull/8609 disabled the use of native memcpy2d backend calls in the L0 plugin due to a bug for non-USM pointers. This commit reenables it.
Goal is to improve compilation time. Store all properties as map. A map is type-list of type-lists. The first entry of the inner list is the property key and the...
Swizzles should not expose mutating functions when the underlying vector is const. This commit SFINAEs these out. This is built on top of https://github.com/intel/llvm/pull/13012.
In https://github.com/intel/llvm/pull/12682 the mutating operators for swizzles (+=, -=, ..., ++, --) were reverted to be members rather than friends. Since swizzles mutate the underlying vec rather than themselves these...
UR PR: https://github.com/oneapi-src/unified-runtime/pull/1385
I've opened a PR in UR that fixes an issue where USM allocation functions are returning non-null pointer values when called with alignment values that are not powers of 2...
So that E2E tests could be updated to include the extensions used explicitly.
Remove extension code from core header files to prevent core bloating and improve compilation times
### Is your feature request related to a problem? Please describe This PR implements the ext_oneapi_forward_progress extension: https://github.com/intel/llvm/pull/13389. This issue concerns the changes made by this PR to the `handler.hpp`...
