Documentation
¶
Overview ¶
Package webgpu implements the WebGPU backend for GPU-accelerated tensor operations. Uses go-webgpu (github.com/go-webgpu/webgpu) for zero-CGO WebGPU bindings.
Package webgpu provides embedded WGSL compute shaders for tensor operations.
Index ¶
- func IsAvailable() (available bool)
- func ListAdapters() (adapters []*wgpu.AdapterInfoGo, err error)
- type Backend
- func (b *Backend) AdapterInfo() *wgpu.AdapterInfoGo
- func (b *Backend) Add(a, other *tensor.RawTensor) *tensor.RawTensor
- func (b *Backend) Conv2D(_, _ *tensor.RawTensor, _, _ int) *tensor.RawTensor
- func (b *Backend) Device() tensor.Device
- func (b *Backend) Div(a, other *tensor.RawTensor) *tensor.RawTensor
- func (b *Backend) MatMul(a, other *tensor.RawTensor) *tensor.RawTensor
- func (b *Backend) MaxPool2D(_ *tensor.RawTensor, _, _ int) *tensor.RawTensor
- func (b *Backend) MemoryStats() MemoryStats
- func (b *Backend) Mul(a, other *tensor.RawTensor) *tensor.RawTensor
- func (b *Backend) Name() string
- func (b *Backend) ReLU(x *tensor.RawTensor) *tensor.RawTensor
- func (b *Backend) Release()
- func (b *Backend) Reshape(t *tensor.RawTensor, newShape tensor.Shape) *tensor.RawTensor
- func (b *Backend) Sigmoid(x *tensor.RawTensor) *tensor.RawTensor
- func (b *Backend) Softmax(x *tensor.RawTensor) *tensor.RawTensor
- func (b *Backend) Sub(a, other *tensor.RawTensor) *tensor.RawTensor
- func (b *Backend) Tanh(x *tensor.RawTensor) *tensor.RawTensor
- func (b *Backend) Transpose(t *tensor.RawTensor, axes ...int) *tensor.RawTensor
- type BufferPool
- type BufferSize
- type MemoryStats
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func IsAvailable ¶
func IsAvailable() (available bool)
IsAvailable checks if WebGPU is available on this system.
func ListAdapters ¶
func ListAdapters() (adapters []*wgpu.AdapterInfoGo, err error)
ListAdapters returns information about all available GPU adapters.
Types ¶
type Backend ¶
type Backend struct {
// contains filtered or unexported fields
}
Backend implements tensor operations on GPU using WebGPU.
func New ¶
New creates a new WebGPU backend. Returns an error if WebGPU is not available or initialization fails.
func (*Backend) AdapterInfo ¶
func (b *Backend) AdapterInfo() *wgpu.AdapterInfoGo
AdapterInfo returns information about the GPU adapter.
func (*Backend) Conv2D ¶
Conv2D performs 2D convolution on GPU. TODO: Implement WGSL compute shader for convolution.
func (*Backend) MaxPool2D ¶
MaxPool2D performs 2D max pooling on GPU. TODO: Implement WGSL compute shader for max pooling.
func (*Backend) MemoryStats ¶
func (b *Backend) MemoryStats() MemoryStats
MemoryStats returns current GPU memory usage statistics.
func (*Backend) Release ¶
func (b *Backend) Release()
Release releases all WebGPU resources. Must be called when the backend is no longer needed.
func (*Backend) Reshape ¶
Reshape returns a tensor with new shape. This is typically a metadata-only operation (zero-copy).
func (*Backend) Softmax ¶
Softmax applies softmax along the last dimension. Expects 2D input [batch_size, num_classes].
type BufferPool ¶
type BufferPool struct {
// contains filtered or unexported fields
}
BufferPool manages GPU buffer reuse to reduce allocation overhead. Buffers are categorized by size and usage flags.
func NewBufferPool ¶
func NewBufferPool(device *wgpu.Device) *BufferPool
NewBufferPool creates a new buffer pool for the given device.
func (*BufferPool) Acquire ¶
func (p *BufferPool) Acquire(size uint64, usage wgpu.BufferUsage) *wgpu.Buffer
Acquire gets a buffer from the pool or creates a new one. Returns a buffer that matches or exceeds the requested size and usage.
func (*BufferPool) Clear ¶
func (p *BufferPool) Clear()
Clear releases all pooled buffers. Should be called when the backend is released.
func (*BufferPool) Release ¶
func (p *BufferPool) Release(buffer *wgpu.Buffer, size uint64, usage wgpu.BufferUsage)
Release returns a buffer to the pool for reuse. If the pool is full, the buffer is immediately released.
func (*BufferPool) Stats ¶
func (p *BufferPool) Stats() (allocated, released, hits, misses uint64, pooledCount int)
Stats returns statistics about buffer pool usage.
type BufferSize ¶
type BufferSize int
BufferSize represents different buffer size categories for pooling.
const ( // SmallBuffer for tensors < 4KB. SmallBuffer BufferSize = iota // MediumBuffer for tensors 4KB-1MB. MediumBuffer // LargeBuffer for tensors > 1MB. LargeBuffer )
type MemoryStats ¶
type MemoryStats struct {
// Total bytes allocated since backend creation
TotalAllocatedBytes uint64
// Peak memory usage in bytes
PeakMemoryBytes uint64
// Number of currently active buffers
ActiveBuffers int64
// Buffer pool statistics
PoolAllocated uint64
PoolReleased uint64
PoolHits uint64
PoolMisses uint64
PooledBuffers int
}
MemoryStats represents GPU memory usage statistics.
Source Files