CUDA Device Management

group cuda_device_management

Functions

inline cuda_device_id get_current_cuda_device()

Returns a cuda_device_id for the current device.

The current device is the device on which the calling thread executes device code.

Returns:

cuda_device_id for the current device

inline int get_num_cuda_devices()

Returns the number of CUDA devices in the system.

Returns:

Number of CUDA devices in the system

inline std::pair<std::size_t, std::size_t> available_device_memory()

Returns the available and total device memory in bytes for the current device.

Returns:

The available and total device memory in bytes for the current device as a std::pair.

inline std::size_t percent_of_free_device_memory(int percent)

Returns the approximate specified percent of available device memory on the current CUDA device, aligned (down) to the nearest CUDA allocation size.

Parameters:

percent – The percent of free memory to return.

Returns:

The recommended initial device memory pool size in bytes.

struct cuda_device_id

#include <cuda_device.hpp>

Strong type for a CUDA device identifier.

Public Types

using value_type = int

Integer type used for device identifier.

Public Functions

inline cuda_device_id() noexcept

Construct a cuda_device_id from the current device.

inline explicit constexpr cuda_device_id(value_type dev_id) noexcept

Construct a cuda_device_id from the specified integer value.

Parameters:

dev_id – The device’s integer identifier

inline constexpr value_type value() const noexcept

The wrapped integer value.

Returns:

The wrapped integer value

struct cuda_set_device_raii

#include <cuda_device.hpp>

RAII class that sets the current CUDA device to the specified device on construction and restores the previous device on destruction.

Public Functions

inline explicit cuda_set_device_raii(cuda_device_id dev_id)

Construct a new cuda_set_device_raii object and sets the current CUDA device to dev_id

Parameters:

dev_id – The device to set as the current CUDA device

inline ~cuda_set_device_raii() noexcept

Reactivates the previous CUDA device.