Steffen Larsen

> Have you seen this #12786 @steffenlarsen? I missed that! Thanks for pointing it out. 😄

> The last comment that needs to be resolved before this PR can get merged is the one at "[#8647 (comment)](https://github.com/intel/llvm/pull/8647#discussion_r1484301864)". > > I'm not sure I fully understand what...

I think this is a reasonable idea, but maybe we should just allow both? The reason we use `acc` is because it is one of the 4 device types in...

For vec and swizzle this is addressed in https://github.com/intel/llvm/pull/12396.

These have been addressed: - `vec`: https://github.com/intel/llvm/pull/12396 - Swizzles: https://github.com/intel/llvm/pull/13012 - `annotated_arg`: https://github.com/intel/llvm/pull/12232 @rolandschulz - I will close this. Please reopen if you believe we missed something.

> I've not worked much with non-GPU targets, so please correct me if I'm wrong here, the reason for padding only for AMD/NVIDIA targets is that those two are the...

@intel/dpcpp-tools-reviewers approval is needed.

Post-commit failure should be addressed by https://github.com/intel/llvm/pull/13889.

These changes are ready for review despite being in draft.

UR changes have been merged (though a small fix patch https://github.com/oneapi-src/unified-runtime/pull/1588 is needed) so this is ready for review. Tag @intel/llvm-reviewers-cuda @intel/dpcpp-tools-reviewers @intel/dpcpp-nativecpu-pi-reviewers @intel/unified-runtime-reviewers