Skip to content

gpuflow.md

Latest commit

 

History

History
78 lines (68 loc) · 2.97 KB

gpuflow.md

File metadata and controls

78 lines (68 loc) · 2.97 KB

GPU.cpp Lifecycle

flowchart TD
  %% Data Preparation & Upload
  subgraph "Data Preparation & Upload"
    A["CPU Data"]
    B["Define Data Properties<br>(shape, type, size)"]
    C["Create GPU Buffer<br>(allocate raw buffer)"]
    D["Create Tensor<br>(allocates Array with one<br> or more buffers<br>and associates Shape)"]
    
    E["Upload Data via toGPU <br>(raw buffer)<br>toGPU<br>(ctx, data, buffer, size)"]
    F["Upload Data via toGPU<br>(Tensor overload)<br>toGPU(ctx, data, tensor)"]
    G["Optional: <br> Kernel Parameters<br>toGPU(ctx, params, Kernel)"]
  end

  %% Buffer Setup & Bindings
  subgraph "Buffer & Binding Setup"
    H["Define Bindings<br>(Bindings, TensorView)"]
    I["Map GPU buffers<br> to shader bindings<br>(Collection from Tensor<br> or single buffers)"]
  end

  %% Kernel Setup & Execution
  subgraph "Kernel Setup & Execution"
    J["Define KernelCode<br>(WGSL template, workgroup size, precision)"]
    K["Create Kernel"]
    L["Dispatch Kernel"]
  end

  %% GPU Execution & Result Readback
  subgraph "GPU Execution & Result Readback"
    M["Kernel Execution<br>(GPU shader runs)"]
    N["Readback Data<br>(toCPU variants)"]
  end

  %% Context & Resources
  O["Context<br>(Device, Queue,<br>TensorPool, KernelPool)"]

  %% Flow Connections
  A --> B
  B --> C
  B --> D
  C --> E
  D --> F
  F --> H
  E --> H
  H --> I
  I --> K
  J --> K
  G --- K
  K --> L
  L --> M
  M --> N

  %% Context shared by all stages
  O --- D
  O --- E
  O --- F
  O --- K
  O --- L
  O --- N
Loading

• The gpu::Array (which wraps a GPU buffer with usage and size) and the gpu::Shape (which defines dimensions and rank) are combined—via the creation process—to produce a gpu::Tensor. • A gpu::TensorView provides a non‑owning view into a slice of a gpu::Tensor. Ex. TensorView view = {tensor, 0, 256}; • gpu::Bindings collect multiple Tensors (or TensorViews) along with view offset/size information for use in a kernel.
• The gpu::TensorPool (managed by the Context) is responsible for the lifetime of tensors and GPU resource cleanup. • gpu::KernelCode contains the WGSL shader template plus metadata (workgroup size, precision, label, and entry point) that drive the kernel configuration.
• The gpu::createKernelAsync/gpu::createKernel functions (within the Execution Flow) use the gpu::Context, gpu::Bindings, and gpu::KernelCode to configure and construct a gpu::Kernel that manages all the underlying GPU resources (buffers, bind groups, compute pipeline, etc.).
• gpu::KernelCode’s workgroup size (a gpu::Shape) defines the dispatch configuration, and the gpu::Kernel eventually uses the underlying gpu::Array (contains WGPUBuffer, WGPUBufferUsage, size_t) and gpu::Shape data from the created Tensor.

gpu::Tensor Ranks: Rank 0: Scalar Rank 1: Vector Rank 2: Matrix Rank 3: 3D Tensor (or Cube) Rank 4: 4D Tensor Rank (max 8): Higher Dimensional Tensors