Steven Perron

Here is a smaller test case that demonstrates the problem: https://godbolt.org/z/hzG9casr1

As with other RawBuffer related issues, we will not be updating that feature. We will leave the layout as it for now. However, we do need to carefully consider the...

I opened https://github.com/microsoft/hlsl-specs/issues/306 to follow up on the vk::BufferPointer layout rules.

This seems like an issue with the layout rules not getting passed around correctly for the vk::SpirvType. Here is another case that fails. It fails differently, but it probably a...

I'm starting to understand the problem, and I'll need to think about how to fix it. The problem in [the smaller example](https://godbolt.org/z/3vq3xWY7v) is that the SPIR-V backend does not know...

As I have thought about this more, DXC does not know the correct layout for the result of a `vk::ext_instruction` function. It cannot generate code that does to or from...

We are able to deal with workgroup pointers, and we are adding this to the HLSL standard headers: https://github.com/microsoft/DirectXShaderCompiler/pull/6873. I am coming to believe that the layout issue with storage...

Opcopylogical only helps a little. It would not translate the data pointed to by a pointer. You would need a load. It also doesn't help with the result type for...

> Its been over a year since I've checkd, but seems that still, contrary to the docs > > ``` > struct Test > { > // The vk::ext_decorate don't...

It is not currently possible to access this builtin. We are designing a change the inline spir-v to make it possible. See https://github.com/microsoft/hlsl-specs/pull/59.