Skip to content

Conversation

aelovikov-intel
Copy link
Contributor

Highlights:

  • No <exception> for device compilation
  • limit half/bfloat16 to forward declarations
  • is_device_copyable.hpp is lightweight but it triggers an FE bug when using PCH for device code, so I limited it to forward declaration for now too.
  • Remove duplicate/redundant includes

Highlights:

* No `<exception>` for device compilation
* limit `half`/`bfloat16` to forward declarations
* `is_device_copyable.hpp` is lightweight but it triggers an FE bug when
  using PCH for device code, so I limited it to forward declaration for
  now too.
* Remove duplicate/redundant includes
typedef __fp16 h2 ATTRIBUTE_EXT_VEC_TYPE(2);
auto rv = __nvvm_ldg_h2(reinterpret_cast<const h2 *>(ptr));
sycl::vec<half, 2> ret;
T ret;
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just makes this statement template-type-dependent, so we don't need a definition of half unless the caller used it (and hence already had it).

Copy link
Contributor

@intel/llvm-gatekeepers please consider merging

@aelovikov-intel aelovikov-intel merged commit fb8d4ab into intel:sycl Sep 25, 2025
47 of 50 checks passed
@aelovikov-intel aelovikov-intel deleted the marray-vec-includes branch September 25, 2025 13:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants