Struct WgpuPipeline 

pub struct WgpuPipeline<'a> { /* private fields */ }

Implementations

impl<'a> WgpuPipeline<'a>

pub fn compile( context: &'a WgpuComputeContext, source: &str, entry_point: &str, ) -> Result<Self, ForgeError>

pub fn compile_spirv( context: &'a WgpuComputeContext, spirv: &[u8], entry_point: &str, ) -> Result<Self, ForgeError>

Like [compile] but accepts pre-compiled SPIR-V bytes (from HLSL→DXC).

pub fn compile_msl( context: &'a WgpuComputeContext, _source: &str, _entry_point: &str, ) -> Result<Self, ForgeError>

Compile pre-emitted MSL source through wgpu’s Metal backend.

wgpu does not expose ShaderSource::Msl — on macOS, wgpu transpiles WGSL→MSL internally via naga. For pre-emitted MSL, the caller should use a native Metal path (e.g. metal-rs). This method exists so the runtime can attempt MSL compilation on macOS; on other platforms it returns an error and the runtime falls back to WGSL.

pub fn dispatch_rayprobe( &self, tlas: &Tlas, buffers: &[BufferView], schedule: &Schedule, element_count: usize, ) -> Result<u64, ForgeError>

Dispatch a ray-query kernel, binding tlas as the acceleration_structure at binding 0 and the supplied buffer views (rays at binding 1, hits at binding 2) at their own bindings. The generic QualiaCompute::dispatch only binds buffers, so this is the dedicated path for the acceleration- structure binding. Returns wall-clock nanoseconds (ray-query passes skip the timestamp path). The caller must keep the TLAS (and its BLAS) alive.

Trait Implementations

impl<'a> QualiaCompute for WgpuPipeline<'a>

fn dispatch( &self, buffers: &[BufferView], schedule: &Schedule, element_count: usize, ) -> Result<u64, ForgeError>

Dispatches a kernel using the supplied Schedule and buffer views.