Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/garciat/wgpu-memory
A utility library for WebGPU that provides strongly-typed ArrayBuffer memory access that is compatible with WGSL's alignment and size specifications.
https://github.com/garciat/wgpu-memory
showcase-project
Last synced: about 11 hours ago
JSON representation
A utility library for WebGPU that provides strongly-typed ArrayBuffer memory access that is compatible with WGSL's alignment and size specifications.
- Host: GitHub
- URL: https://github.com/garciat/wgpu-memory
- Owner: Garciat
- License: mit
- Created: 2024-12-24T08:27:12.000Z (about 2 months ago)
- Default Branch: main
- Last Pushed: 2025-02-04T20:59:53.000Z (8 days ago)
- Last Synced: 2025-02-04T21:40:37.657Z (8 days ago)
- Topics: showcase-project
- Language: TypeScript
- Homepage: https://garciat.com/wgpu-memory/
- Size: 223 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# wgpu-memory
[](https://jsr.io/@garciat/wgpu-memory)
[](https://jsr.io/@garciat/wgpu-memory)
[](https://github.com/garciat/wgpu-memory)
[](https://codecov.io/gh/garciat/wgpu-memory)
A utility library for WebGPU that provides strongly-typed `ArrayBuffer` memory
access that is compatible with [WGSL](https://gpuweb.github.io/gpuweb/wgsl/)'s
[alignment and size specifications](https://gpuweb.github.io/gpuweb/wgsl/#alignment-and-size).
Supported types are:
- [Scalar Types](https://gpuweb.github.io/gpuweb/wgsl/#scalar-types)
- `bool`
- `i32`
- `u32`
- `f32`
- `f16` - iff [Float16](https://github.com/tc39/proposal-float16array) is
implemented by the environment.
- [Vector Types](https://gpuweb.github.io/gpuweb/wgsl/#vector-types)
- `vec2`
- `vec3`
- `vec4`
- [Matrix Types](https://gpuweb.github.io/gpuweb/wgsl/#matrix-types)
- `mat2x2`
- `mat3x3`
- `mat4x4`
- [Array Types](https://gpuweb.github.io/gpuweb/wgsl/#array-types)
- `array` (fixed-size arrays)
- [Structure Types](https://gpuweb.github.io/gpuweb/wgsl/#struct-types)
- `struct { M_1 : T_1 , ... , M_N : T_N }`
NOTE: if you're looking for a WebGPU-compatible math library, I recommend
[greggman/wgpu-matrix](https://github.com/greggman/wgpu-matrix). It also works
directly on `ArrayBuffer`s without additional object wrapping.
## Example
Importing the library:
```typescript
import * as memory from "jsr:@garciat/wgpu-memory";
```
Defining types:
```typescript
const Vertex = memory.StructOf({
position: { index: 0, type: memory.Vec4F },
color: { index: 1, type: memory.Vec4F },
normal: { index: 2, type: memory.Vec3F },
uv: { index: 3, type: memory.Vec2F },
});
const VertexQuad = memory.ArrayOf(Vertex, 6);
const CubeMesh = memory.ArrayOf(VertexQuad, 6);
const Instance = memory.StructOf({
tint: { index: 0, type: memory.Vec4F },
model: { index: 1, type: memory.Mat4x4F },
mvMatrix: { index: 2, type: memory.Mat4x4F },
normalMatrix: { index: 3, type: memory.Mat4x4F },
});
```
Using the types to allocate and write data:
```typescript
import { mat4, vec3 } from "npm:[email protected]";
const cubeMeshData = memory.allocate(CubeMesh);
{
const view = new DataView(cubeMeshData);
// The input is type-checked to match the required nested shape of the data
CubeMesh.write(view, [
// Front face
// deno-fmt-ignore
[
{ position: [-1, -1, 1, 1], color: [1, 0, 0, 1], normal: [0, 0, 1], uv: [0, 0] },
{ position: [1, -1, 1, 1], color: [0, 1, 0, 1], normal: [0, 0, 1], uv: [1, 0] },
{ position: [1, 1, 1, 1], color: [0, 0, 1, 1], normal: [0, 0, 1], uv: [1, 1] },
{ position: [-1, 1, 1, 1], color: [1, 1, 1, 1], normal: [0, 0, 1], uv: [0, 1] },
{ position: [-1, -1, 1, 1], color: [1, 0, 0, 1], normal: [0, 0, 1], uv: [0, 0] },
{ position: [1, 1, 1, 1], color: [0, 0, 1, 1], normal: [0, 0, 1], uv: [1, 1] },
],
// etc
]);
}
const cubeInstanceData = memory.allocate(Instance, 2);
{
const view = new DataView(cubeInstanceData);
{
const position = vec3.fromValues(0, 0, 0);
const scale = vec3.fromValues(0.5, 0.5, 0.5);
// Write into the 0th instance's `tint` field
Instance.fields.tint.writeAt(view, 0, [1, 1, 1, 1]);
// Get a Float32Array view into the 0th instance's `model` field
const model = Instance.fields.model.viewAt(cubeInstanceData, 0);
// This view can be directly used with the wgpu-matrix library
// No copying needed at all
mat4.identity(model);
mat4.translate(model, position, model);
mat4.scale(model, scale, model);
}
{
const position = vec3.fromValues(0, 0, -3);
const scale = vec3.fromValues(0.05, 0.05, 0.05);
Instance.fields.tint.writeAt(view, 1, [1, 1, 1, 1]);
const model = Instance.fields.model.viewAt(cubeInstanceData, 1);
mat4.identity(model);
mat4.translate(model, position, model);
mat4.scale(model, scale, model);
}
}
```