Function csr_to_csc 

pub fn csr_to_csc<T: DataTypeLike>(
    ctx: &Context,
    row_count: usize,
    col_count: usize,
    nonzero_count: usize,
    csr_values: &DeviceMemory<T>,
    csr_row_offsets: &DeviceMemory<i32>,
    csr_column_indices: &DeviceMemory<i32>,
    csc_values: &mut DeviceMemory<T>,
    csc_column_offsets: &mut DeviceMemory<i32>,
    csc_row_indices: &mut DeviceMemory<i32>,
    action: Action,
    index_base: IndexBase,
    algorithm: CsrToCscAlgorithm,
    external_buffer: &mut DeviceMemory<u8>,
) -> Result<()>

Converts a sparse matrix in CSR format into a sparse matrix in CSC format. The resulting matrix can also be seen as the transpose of the original sparse matrix. This can also convert a matrix in CSC format into a matrix in CSR format.

This operation requires extra storage proportional to nonzero_count. It always produces the same matrix values in the destination format.

The operation executes asynchronously with respect to the host and may return before the result is ready.

csr_to_csc_buffer_size returns the size of the workspace needed by csr_to_csc. Pass a buffer of this size to csr_to_csc.

If nonzero_count is 0, then the CSR column indices, CSR values, CSC values, and CSC row indices may be zero-length buffers. In this case, all CSC column pointers are initialized to the configured index base.

If m or n is 0, the pointers are not checked and the operation succeeds with Ok.

csr_to_csc supports the following data types:

X/Y	Notes
DataType::I8
DataType::F16
DataType::Bf16
DataType::F32
DataType::F64
DataType::ComplexF16	Deprecated.
DataType::ComplexBf16	Deprecated.
DataType::ComplexF32
DataType::ComplexF64

csr_to_csc supports the following algorithms (CsrToCscAlgorithm):

Algorithm	Notes
CsrToCscAlgorithm::Default	Default algorithm.

Action	Notes
Action::Symbolic	Compute the structure of the CSC output matrix (offset, row indices).
Action::Numeric	Compute the structure of the CSC output matrix and copy the values.

csr_to_csc has the following properties:

• Requires no extra storage.
• Supports asynchronous execution.

csr_to_csc supports the following optimizations:

• CUDA graph capture.
• Hardware Memory Compression.