Two-row matrices in uniform buffers in Vulkan/SPIR-V's default
"extended" (std140) layout do not adhere to WGSL/Naga IR's layout
rules - each column is aligned to a minimum of 16 bytes rather than
just the vector size.
To work around this, we emit additional std140 compatible type
declarations for each type used in a uniform buffer that requires
one. Any two-row matrix struct member is decomposed into the
containing struct as separate vectors for each column. Two-row
matrices used directly as uniform buffers are wrapped in a struct with
a vector member for each column, and arrays (of arrays, etc) of
two-row matrices are declared as arrays (of arrays) of structs
containing a vector member for each column.
When loading such a value from a uniform buffer, we convert from the
std140 compatible type to the regular type immediately after loading.
Accesses of a uniform two-row matrix's columns using constant indices
are rewritten to access the containing struct's vector members
directly. Accesses using dynamic indices are implemented by loading
and converting the matrix, extracting each column then `switch`ing on
the index to select the correct column.
We can now remove the expected failure annotation for the Vulkan
backend on the uniform_input struct_layout test, as it now passes.
This means that usage of a `StagingBelt` only requires passing along
the current `CommandEncoder` instead of both the encoder and the device.
It also makes the API reflect the fact that a `StagingBelt` cannot be
reused with a fresh device.
As far as I know, the only reason this was not done previously was that
`Device` was not clonable, and so doing this would have required an
`Arc<Device>` or other specific mechanism.
The plumbing to get it out of `wgpu_core` is awkward but I didn’t see
a clearly better option. I think it would make sense to have something
like `Global::get_label<T>(id: Id<T>) -> String` but that looks like a
lot more work.
* Add documentation that every `allocate()` or `write_buffer()`
operation must have a size that is a multiple of 4.
* Add assertions for those properties (this gives more helpful panics
than leaving it to validation).
* Add a randomized test to exercise usage of StagingBelt.
I was concerned that `StagingBelt` might have an alignment math bug;
having written this test, I am much less concerned. There was previously
no test for `StagingBelt` at all, if you don’t count examples/skybox.
* Removed unnecessary dependency in `wgpu` on `hashbrown`.
* Removed unused dependencies in `naga-test`.
* Moved some dependencies in `wgpu-test` to `[dev-dependencies]`
(see comment for why this is worth doing).
Candidates were discovered by `cargo clippy -- -Wunused-crate-dependencies`.
This command reports whether each dependency is unused in each *crate*,
so it has false positives when a dependency is used by only some crates
in a package, but it is still useful as a short-list.
* Features and draw commands added
* Tried to implement the pipeline creation (completely untested)
* Fixed clippy issues
* Fixed something I think
* A little bit of work on the mesh shader example (currently doesn't work on dx12)
* Reached a new kind of error state
* Fixed an alignment issue
* DirectX 12 mesh shaders working :party:
* Removed stupid change and updated changelog
* Fixed typo
* Added backends option to example framework
* Removed silly no write fragment shader from tests to see if anything breaks
* Tried to make mesh shader tests run elsewhere too
* Removed printlns and checked that dx12 mesh shader tests run
* Documented very strange issue
* I'm so lost
* Fixed stupid typos
* Fixed all issues
* Removed unnecessary example stuff, updated tests
* Updated typos.toml
* Updated limits
* Apply suggestion from @cwfitzgerald
Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
* Apply suggestion from @cwfitzgerald
Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
* Removed supported backends, made example & tests always pass the filename to shader compilers
* Removed excessive bools in test params
* Added new tests to the list
* I'm a sinner for this one (unused import)
* Replaced random stuff with test params hashing
* Updated typos.toml
* Updated -Fo typo thing
* Actually fixed typo issue this time
* Update CHANGELOG.md
Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
* Update tests/tests/wgpu-gpu/mesh_shader/mod.rs
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Update wgpu-hal/src/dx12/mod.rs
Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
* Addressed comments
* Lmao
---------
Co-authored-by: Connor Fitzgerald <connorwadefitzgerald@gmail.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>