WithBinding

Extends

Withable<WithBinding>

Extended by

TgpuRoot

Methods

createComputePipeline()

createComputePipeline<ComputeIn>(descriptor): TgpuComputePipeline

Type Parameters

• ComputeIn extends IORecord<AnyComputeBuiltin>

Parameters

descriptor

Descriptor<ComputeIn>

Returns

TgpuComputePipeline

Defined in

packages/typegpu/src/core/root/rootTypes.ts:253

createGuardedComputePipeline()

createGuardedComputePipeline<TArgs>(callback): TgpuGuardedComputePipeline<TArgs>

Creates a compute pipeline that executes the given callback in an exact number of threads. This is different from withCompute(...).createPipeline() in that it does a bounds check on the thread id, where as regular pipelines do not and work in units of workgroups.

Type Parameters

• TArgs extends number[]

Parameters

callback

(…args) => void

A function converted to WGSL and executed on the GPU. It can accept up to 3 parameters (x, y, z) which correspond to the global invocation ID of the executing thread.

Returns

TgpuGuardedComputePipeline<TArgs>

Examples

If no parameters are provided, the callback will be executed once, in a single thread.

const fooPipeline = root
  .createGuardedComputePipeline(() => {
    'use gpu';
    console.log('Hello, GPU!');
  });

fooPipeline.dispatchThreads();
// [GPU] Hello, GPU!

One parameter means n-threads will be executed in parallel.

const fooPipeline = root
  .createGuardedComputePipeline((x) => {
    'use gpu';
    if (x % 16 === 0) {
      // Logging every 16th thread
      console.log('I am the', x, 'thread');
    }
  });

// executing 512 threads
fooPipeline.dispatchThreads(512);
// [GPU] I am the 256 thread
// [GPU] I am the 272 thread
// ... (30 hidden logs)
// [GPU] I am the 16 thread
// [GPU] I am the 240 thread

Defined in

packages/typegpu/src/core/root/rootTypes.ts:417

createRenderPipeline()

Call Signature

createRenderPipeline<TVertexIn, TAttribs, TVertexOut, TFragmentOut>(descriptor): TgpuRenderPipeline<NormalizeOutput<TFragmentOut>>

Type Parameters

• TVertexIn extends In = Record<string, any>

• TAttribs extends object = LayoutToAllowedAttribs<TVertexIn>

• TVertexOut = unknown

• TFragmentOut = unknown

Parameters

descriptor

DescriptorBase & object

Returns

TgpuRenderPipeline<NormalizeOutput<TFragmentOut>>

Defined in

packages/typegpu/src/core/root/rootTypes.ts:257

Call Signature

createRenderPipeline<TVertexIn, TAttribs, TVertexOut>(descriptor): TgpuRenderPipeline<Void>

Type Parameters

• TVertexIn extends In = Record<string, any>

• TAttribs extends object = LayoutToAllowedAttribs<TVertexIn>

• TVertexOut extends Out = Out

Parameters

Call Signature

createRenderPipeline<TVertexIn, TAttribs, TVertexOut, TFragmentOut>(descriptor): TgpuRenderPipeline<Void> | TgpuRenderPipeline<NormalizeOutput<TFragmentOut>>

Type Parameters

• TVertexIn extends In = Record<string, any>

• TAttribs extends object = LayoutToAllowedAttribs<TVertexIn>

• TVertexOut extends Out = Out

• TFragmentOut = unknown

Parameters

descriptor

DescriptorBase & {attribs: TAttribs;fragment: TgpuFragmentFn<VertexOutToVarying<OmitBuiltins<TVertexOut>> & Record<string, AnyFragmentInputBuiltin>, Assume<TFragmentOut, Out>> | (input) => AutoFragmentOut<Assume<TFragmentOut, v4f | AnyAutoCustoms>>;targets: FragmentOutToTargets<NoInfer<TFragmentOut>>;vertex: TgpuVertexFn<TVertexIn, Assume<TVertexOut, Out>> | (input) => AutoVertexOut<Assume<TVertexOut, AnyAutoCustoms>>; } | {attribs: TAttribs;fragment: TgpuFragmentFn<VertexOutToVarying<OmitBuiltins<TVertexOut>> & Record<string, AnyFragmentInputBuiltin>, Record<string, never>> | (input) => undefined;targets: undefined;vertex: TgpuVertexFn<TVertexIn, Assume<TVertexOut, Out>> | (input) => AutoVertexOut<Assume<TVertexOut, AnyAutoCustoms>>; }