alan-baker

c11_atomics: Filter invalid memory orders for 2 order instructions

2

* OpenCL C (and SPIR-V) require that the failure memory order is not stronger than the success memory order. Also see Khronos internal memory model issue #181 CC @bashbaug

Add CTS for texture_and_sampler_let language feature

1

CTS for https://github.com/gpuweb/gpuweb/pull/5389 Issue: # **Requirements for PR author:** - [x] All missing test coverage is tracked with "TODO" or `.unimplemented()`. - [x] New helpers are `/** documented */` and...

Compile and link flows lose compile options

3

When performing compile and link flows clspv is instructed to compile to IR. This skips the optimizations passes; however currently clspv needs to know several of those options to compile...

[SPV_KHR_fma](https://github.com/KhronosGroup/SPIRV-Registry/pull/361) adds a new `OpFmaKHR` instruction for correctly rounded fma. There is a corresponding Vulkan extension that allows implementations to advertise a correctly rounded fma. This should be preferred over...

[buffer-view] Extent of reinterpretation permitted

1

Proposal currently limits reinterpretation to always from the base variable and only once. Would more extensive reinterpretation be better? This seems it would imply taking any pointer type and reinterpreting...

The proposal currently says the size must be an override-expression, but storage and uniform require a const-expression (storage can additionally be unsized). So only workgroup variables allow an override-sized buffer....

Add subgroup_uniformity language extension

2

* Adds subgroup uniformity language * When supported uniformity analysis is performed at twice (at different scopes) See: #5368

[subgroup matrix] Allow subgroup_matrix types in private address space?

3

The proposal describes only allowing subgroup_matrix types in Function or Private address spaces. Private is a little problematic because of the subgroup uniformity analysis. All mutable global variables are considered...

Allow lets of textures and samplers

4

Refs #5339 * Add a language feature, `texture_and_sampler_let`, that allows the effective value type of a let to be a texture or sampler type

Should we not implement `bufferLength` and instead deprecate `arrayLength` and introduce a single `length` function. Note that we already have a geometric length function so maybe that name becomes too...