ExecutionHybrid#

class nvmath.sparse.advanced.ExecutionHybrid(device_id: int = 0, num_threads: int | None = None)[source]#

A data class for providing hybrid (CPU-GPU) execution options to the DirectSolver object and the wrapper function direct_solver().

device_id#

CUDA device ordinal (only used if the operand resides on the CPU). The default value is 0.

Type:

int

num_threads#

The number of CPU threads used to execute the plan. If not specified, defaults to the number of CPU cores available to the process.

Type:

int | None

See also

ExecutionCUDA, DirectSolver, direct_solver().