cuda.core.Kernel#

class cuda.core.Kernel(*args, **kwargs)#

Represent a compiled kernel that had been loaded onto the device.

Kernel instances can execution when passed directly into the launch() function.

Directly creating a Kernel is not supported, and they should instead be created through a ObjectCode object.

Methods

__init__(*args, **kwargs)#
static from_handle(
handle,
ObjectCode mod: ObjectCode = None,
) → Kernel#

Creates a new Kernel object from a kernel handle.

Parameters:

  • handle (int) – Kernel handle representing the address of a foreign kernel object (CUkernel).

  • mod (ObjectCode, optional) – The ObjectCode object associated with this kernel. Provides library lifetime for foreign kernels not created by cuda.core.

Attributes

arguments_info#

list[ParamInfo]

list[ParamInfo]: (offset, size) for each argument of this function

Type:

Kernel.arguments_info

attributes#

KernelAttributes

Get the read-only attributes of this kernel.

Type:

Kernel.attributes

handle#

Return the underlying kernel handle object.

Caution

This handle is a Python object. To get the memory address of the underlying C handle, call int(Kernel.handle).

num_arguments#

int

int : The number of arguments of this function

Type:

Kernel.num_arguments

occupancy#

KernelOccupancy

Get the occupancy information for launching this kernel.

Type:

Kernel.occupancy