Struct SparseFFTKernel 

pub struct SparseFFTKernel { /* private fields */ }

Kernel for computing sparse FFT on GPU

Implementations

impl SparseFFTKernel

pub fn new( input_size: usize, sparsity: usize, input_address: usize, output_values_address: usize, output_indices_address: usize, algorithm: SparseFFTAlgorithm, window_function: WindowFunction, ) -> Self

Create a new sparse FFT kernel

pub fn apply_window(&self) -> FFTResult<KernelStats>

Apply window function on GPU.

As with Self::execute, no real device launch happens here, so the returned stats are honest analytical estimates with None for the un-measurable bandwidth. Windowing is an element-wise O(n) multiply, so its modelled FLOP count is n (one multiply per sample).

pub fn get_algorithm_implementation(&self) -> FFTResult<KernelImplementation>

Get algorithm-specific implementation

Trait Implementations

impl Debug for SparseFFTKernel

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl GPUKernel for SparseFFTKernel

fn name(&self) -> &str

Get kernel name

fn config(&self) -> &KernelConfig

Get kernel configuration

fn set_config(&mut self, config: KernelConfig)

Set kernel configuration

fn execute(&self) -> FFTResult<KernelStats>

Execute kernel