Struct CudaMemoryManager 

pub struct CudaMemoryManager { /* private fields */ }

CUDA memory management for sparse matrices

Implementations

impl CudaMemoryManager

pub fn new() -> Self

Create a new CUDA memory manager

pub fn allocate_sparse_matrix<T>( &mut self, _matrix: &CsrArray<T>, _device: &GpuDevice, ) -> Result<CudaMatrixBuffers<T>, GpuError>

where T: GpuDataType + Copy + Float + SparseElement + Debug,

Allocate GPU memory for sparse matrix data

pub fn allocate_optimized<T>( &mut self, _data: &[T], _device: &GpuDevice, _access_pattern: MemoryAccessPattern, ) -> Result<GpuBuffer<T>, GpuError>

where T: GpuDataType + Copy + Float + SparseElement + Debug,

Allocate GPU memory with optimal memory layout

Trait Implementations

impl Default for CudaMemoryManager

fn default() -> Self

Returns the “default value” for a type.