go-webgpu

A WebGPU binding for Go programming language. Based on codegen.

Re-generate codes

go run ./go-webgpu-gen \
    -src webgpu-headers/webgpu.yml \
    -out webgpu \
    -importPath "github.com/Tnze/go-webgpu/webgpu"

Project Configuration

To use this library in your project. Considering how do you want to link to the WebGPU implementation.

There is mainly two ways to call WebGPU APIs supported by this library: By CGO or syscall.LazyDLL/LazyProc (Windows only).

By CGO

By using CGO, you can link WebGPU implementation dynamically or statically. To doing so, you should set the CGO_LDFLAGS before building your Go project, specific the WebGPU implementation library linkage options.

Further more, you can have more flexible controls by using generic LDFLAGS options, e.g., static linking.

For example, on Linux/macOS:

export CGO_ENABLE=1
export CGO_LDFLAGS="-L/path/to/wgpu-native-lib -lwgpu-native"

On Windows:

$env:CC="C:\Program Files\Microsoft Visual Studio\18\Community\VC\Tools\Llvm\x64\bin\clang.exe"
$env:CGO_ENABLE="1"
$env:CGO_LDFLAGS="-L/path/to/wgpu-native-lib -lwgpu-native"

By syscall

On windows, we can call DLLs without enabling CGO, by using syscall std package. In this case, make sure the wgpu-native.dll in the searching path is only thing we need to do.

TODO: make the DLL name configurable.