cuda::link_t Class Reference

Wrapper class for a CUDA link (a process of linking compiled code together into an executable binary, using CUDA, at run-time) More...

#include <link.hpp>

Public Member Functions
device::id_t	device_id () const noexcept
	The raw CUDA ID for the device w.r.t. which the link is defined.
context::handle_t	context_handle () const noexcept
	The raw CUDA handle for the context in which the represented stream is defined.
bool	is_owning () const noexcept
	True if this wrapper is responsible for telling CUDA to destroy the link upon the wrapper's own destruction.
device_t	device () const
	The device w.r.t. which the link is defined.
context_t	context () const
	The context in which this stream was defined.
memory::region_t	complete () const
	Complete a linking process, producing a completely-linked cubin image (for loading into modules). More...
void	add (link::input::image_t image, const link::options_t &ptx_compilation_options={}) const
	Add another linkable image, from memory, to this linking-process. More...
	link_t (const link_t &)=delete
	link_t (link_t &&other) noexcept
link_t &	operator= (const link_t &)=delete
link_t &	operator= (link_t &&other) noexcept
void	add_file (link::input::file_t file_input, const link::options_t &options={}) const
	Add another linkable image, from a file, to this linking-process. More...
void	add_file (const char *path, link::input_kind_t file_contents_type, const link::options_t &options={}) const

Friends
link_t	link::wrap (device::id_t, context::handle_t, link::handle_t, const link::options_t &, bool) noexcept

Detailed Description

Wrapper class for a CUDA link (a process of linking compiled code together into an executable binary, using CUDA, at run-time)

Note

This class is a "reference type", not a "value type". Therefore, making changes to the link is a const-respecting operation on this class.

Member Function Documentation

add()

void cuda::link_t::add ( link::input::image_t  image, const link::options_t &  ptx_compilation_options = {}  ) const

inline

Add another linkable image, from memory, to this linking-process.

Parameters

[in]	image	Memory region containing the image
[in]	ptx_compilation_options	Options for compiling PTX code to cubin, if necessary, before linking.

Note

some types of linkable images are not, in fact, even compiled - but can be compiled by the driver with the specified options.

add_file()

void cuda::link_t::add_file ( link::input::file_t  file_input, const link::options_t &  options = {}  ) const

inline

Add another linkable image, from a file, to this linking-process.

Parameters

[in]	file_input	Path of the image file to be added
[in]	ptx_compilation_options	Options for compiling PTX code to cubin, if necessary, before linking.

Note

some types of linkable images are not, in fact, even compiled - but can be compiled by the driver with the specified options.

complete()

memory::region_t cuda::link_t::complete ( ) const

inline

Complete a linking process, producing a completely-linked cubin image (for loading into modules).

Returns

The completely-linked cubin image, in a sized memory range. This memory is owned by the link object, and must not be freed/deleted.

---

The documentation for this class was generated from the following file:

• src/cuda/api/link.hpp