3fda684eb9
See the comments in the code for details. This patch emits the definition of the macro only when the first loop is encountered. This does make that first loop's code look a bit odd: it would be more natural to define the macro at the top of the file. (See the modified files in `naga/tests/out/msl`.) Rejected alternatives: - We could emit the macro definition unconditionally at the top of the file. But this changes every MSL snapshot output file, whereas only eight of them actually contain loops. - We could have the validator flag modules that contain loops. But the changes end up being not small, and spread across the validator, so this seems disproportionate. If we had other consumers of this information, it might make sense. - We could change the MSL backend to allow text to be generated out of order, so that we can decide whether to define the macro after we've generated all the function bodies. But at the moment this seems like unnecessary complexity, although it might be worth doing in the future if we had additional uses for it - say, to conditionally emit helper function definitions. Fixes #4972. |
||
---|---|---|
.. | ||
.cargo | ||
fuzz | ||
hlsl-snapshots | ||
src | ||
tests | ||
xtask | ||
.gitattributes | ||
.gitignore | ||
build.rs | ||
Cargo.toml | ||
CHANGELOG.md | ||
README.md |
Naga
The shader translation library for the needs of wgpu.
Supported end-points
Front-end | Status | Feature | Notes |
---|---|---|---|
SPIR-V (binary) | ✅ | spv-in | |
WGSL | ✅ | wgsl-in | Fully validated |
GLSL | 🆗 | glsl-in | GLSL 440+ and Vulkan semantics only |
Back-end | Status | Feature | Notes |
---|---|---|---|
SPIR-V | ✅ | spv-out | |
WGSL | 🆗 | wgsl-out | |
Metal | ✅ | msl-out | |
HLSL | ✅ | hlsl-out | Shader Model 5.0+ (DirectX 11+) |
GLSL | 🆗 | glsl-out | GLSL 330+ and GLSL ES 300+ |
AIR | |||
DXIL/DXIR | |||
DXBC | |||
DOT (GraphViz) | 🆗 | dot-out | Not a shading language |
✅ = Primary support — 🆗 = Secondary support — 🚧 = Unsupported, but support in progress
Conversion tool
Naga can be used as a CLI, which allows testing the conversion of different code paths.
First, install naga-cli
from crates.io or directly from GitHub.
# release version
cargo install naga-cli
# development version
cargo install naga-cli --git https://github.com/gfx-rs/wgpu.git
Then, you can run naga
command.
naga my_shader.wgsl # validate only
naga my_shader.spv my_shader.txt # dump the IR module into a file
naga my_shader.spv my_shader.metal --flow-dir flow-dir # convert the SPV to Metal, also dump the SPIR-V flow graph to `flow-dir`
naga my_shader.wgsl my_shader.vert --profile es310 # convert the WGSL to GLSL vertex stage under ES 3.20 profile
As naga includes a default binary target, you can also use cargo run
without installation. This is useful when you develop naga itself or investigate the behavior of naga at a specific commit (e.g. wgpu might pin a different version of naga than the HEAD
of this repository).
cargo run my_shader.wgsl
Development workflow
The main instrument aiding the development is the good old cargo test --all-features --workspace
,
which will run the unit tests and also update all the snapshots. You'll see these
changes in git before committing the code.
If working on a particular front-end or back-end, it may be convenient to
enable the relevant features in Cargo.toml
, e.g.
default = ["spv-out"] #TEMP!
This allows IDE basic checks to report errors there unless your IDE is sufficiently configurable already.
Finally, when changes to the snapshots are made, we should verify that the produced shaders are indeed valid for the target platforms they are compiled for:
cargo xtask validate spv # for Vulkan shaders, requires SPIRV-Tools installed
cargo xtask validate msl # for Metal shaders, requires XCode command-line tools installed
cargo xtask validate glsl # for OpenGL shaders, requires GLSLang installed
cargo xtask validate dot # for dot files, requires GraphViz installed
cargo xtask validate wgsl # for WGSL shaders
cargo xtask validate hlsl dxc # for HLSL shaders via DXC
cargo xtask validate hlsl fxc # for HLSL shaders via FXC