Struct TensorManipulationKernel 

pub struct TensorManipulationKernel {
    pub operation: String,
    pub input_size: usize,
    pub output_size: usize,
    pub complexity_factor: f64,
}

GPU Tensor manipulation async kernel

Fields

operation: String

input_size: usize

output_size: usize

complexity_factor: f64

Implementations

impl TensorManipulationKernel

pub fn new_gather(input_size: usize, indices_size: usize) -> Self

pub fn new_scatter(tensor_size: usize, updates_size: usize) -> Self

pub fn new_roll(size: usize) -> Self

pub fn new_where(_condition_size: usize, total_size: usize) -> Self

Trait Implementations

impl AsyncKernel for TensorManipulationKernel

fn execute(&self) -> Pin<Box<dyn Future<Output = Result<()>> + Send>>

Execute the kernel asynchronously

fn name(&self) -> &str

Get kernel name for debugging

fn estimated_time_us(&self) -> u64

Get estimated execution time in microseconds

fn can_batch_with(&self, other: &dyn AsyncKernel) -> bool

Check if this kernel can be batched with another

fn priority(&self) -> Priority

Get kernel priority

fn supports_concurrency(&self) -> bool

Check if kernel supports concurrent execution

fn memory_requirements(&self) -> usize

Get memory requirements in bytes