Struct EllMatrix 

pub struct EllMatrix<T: GpuFloat> { /* private fields */ }

A sparse matrix in ELLPACK (ELL) format, stored on GPU.

Both col_idx and values have length rows * max_nnz_per_row, stored in column-major order: element (i, k) is at index k * rows + i.

Implementations

impl<T: GpuFloat> EllMatrix<T>

pub fn from_host( rows: u32, cols: u32, max_nnz_per_row: u32, col_idx: &[i32], values: &[T], ) -> SparseResult<Self>

Creates an ELL matrix from host-side padded arrays, uploading to GPU.

Arguments

• rows – Number of rows.
• cols – Number of columns.
• max_nnz_per_row – Maximum entries stored per row.
• col_idx – Padded column indices, length rows * max_nnz_per_row, column-major. Unused entries must be ELL_SENTINEL (-1).
• values – Padded values, length rows * max_nnz_per_row, column-major. Unused entries should be zero.

Errors

Returns SparseError::InvalidFormat if array lengths are incorrect.

pub fn from_csr(csr: &CsrMatrix<T>) -> SparseResult<Self>

Creates an ELL matrix from a CSR matrix on the host.

Determines max_nnz_per_row from the CSR structure, then pads each row to that width.

Errors

Returns SparseError::Cuda on transfer failure.

pub fn to_host(&self) -> SparseResult<(Vec<i32>, Vec<T>)>

Downloads the ELL arrays from GPU to host memory.

Errors

Returns SparseError::Cuda on transfer failure.

pub fn rows(&self) -> u32

Returns the number of rows.

pub fn cols(&self) -> u32

Returns the number of columns.

pub fn max_nnz_per_row(&self) -> u32

Returns the maximum non-zeros per row.

pub fn col_idx(&self) -> &DeviceBuffer<i32>

Returns a reference to the column index device buffer.

pub fn values(&self) -> &DeviceBuffer<T>

Returns a reference to the values device buffer.