Austin Eng

The actual error it is hitting is convoluted to derive from the Vulkan spec. It is: - VUID-VkImageCreateInfo-imageCreateMaxMipLevels-02251 Each of the following values (as described in [Image Creation Limits](https://registry.khronos.org/vulkan/specs/1.3-extensions/html/vkspec.html#resources-image-creation-limits)) must...

To clarify - is the problem that it is not desirable to pass `VK_IMAGE_CREATE_EXTENDED_USAGE_BIT`, or that it's odd to allow "creation" of srgb storage views when binding such a view...

based on missing code coverage, testing should be sure to include: * merging to use max minBufferSize * merging stage visibilities * incompatibilities between entries. like sampleType mismatch or storageTexture...

Some drivers have difficulty with the following. Particularly with depth-stencil textures. We should make sure it has some test coverage: - Making a texture view and then using it (sampling,...

this is in the TODOs in the tests already, but from code cov, we also need to test all dimensions (untested: 1d/3d/cube/cube-array)

I also wrote https://github.com/gpuweb/cts/pull/981 but it's stale now. It kinda is a total overhaul of things so it may take some time to actually land. That is, we don't have...

Furthermore, according to https://renderdoc.org/vkspec_chunked/chap5.html#commandbuffers > Each command buffer manages state independently of other command buffers. There is no inheritance of state across primary and secondary command buffers, or between secondary...

Yup, that would be my conclusion as well. Obviously, an implementation can do other state tracking to avoid deduplicating API calls. ex.) D3D12 state does persist after execution, so you...

Following up from performance discussion in the latest conference call: I likely won't have a chance to do a performance test of the late implicit vkQueueSubmit to do the queue...

I believe for this PR as well, unknown should be value 0x1, and destroyed should be value 0x2