cuda-python の API リファレンス
cuda-python (NVIDIA/cuda-python) の公開 API 53 件 —— クラス 27、関数 18、メソッド 8。実際のソースを静的解析して抽出した正確なシグネチャを掲載しています。
リポジトリ: NVIDIA/cuda-python
| 種別 | 件数 |
|---|---|
| クラス | 27 |
| 関数 | 18 |
| メソッド | 8 |
API 一覧
func
cuda_bindings.cuda.bindings.utils._ptx_utils.get_ptx_ver(ptx:str) -> strExtract the PTX ISA version string from PTX source code.
class
cuda_core.cuda.core._host.HostHost (CPU) location for managed-memory operations.
func
cuda_core.cuda.core._utils.enum_explanations_helpers.clean_enum_member_docstring(doc:str | None) -> str | NoneTurn an enum member ``__doc__`` into plain text.
method
cuda_core.cuda.core.checkpoint.Process.lock(timeout_ms:int=0) -> NoneLock this process, blocking further CUDA API calls.
method
cuda_core.cuda.core.checkpoint.Process.pid() -> intProcess ID of the CUDA process.
method
cuda_core.cuda.core.checkpoint.Process.restore(gpu_mapping:Mapping[Any, Any] | None=None) -> NoneRestore this checkpointed process.
method
cuda_core.cuda.core.checkpoint.Process.restore_thread_id() -> intCUDA restore thread ID for this process.
method
cuda_core.cuda.core.checkpoint.Process.state() -> _ProcessStateTypeCUDA checkpoint state for this process.
class
cuda_core.cuda.core.system.typing.AddressingModeAddressing mode of a device.
class
cuda_core.cuda.core.system.typing.AffinityScopeScope for affinity queries.
class
cuda_core.cuda.core.system.typing.ClockIdClock Ids.
class
cuda_core.cuda.core.system.typing.ClockTypeClock types.
class
cuda_core.cuda.core.system.typing.ClocksEventReasonsReasons for a clocks event.
class
cuda_core.cuda.core.system.typing.CoolerControlCooler control type.
class
cuda_core.cuda.core.system.typing.CoolerTargetCooler target.
class
cuda_core.cuda.core.system.typing.DeviceArchDevice architecture.
class
cuda_core.cuda.core.system.typing.FanControlPolicyFan control policies.
class
cuda_core.cuda.core.system.typing.GpuP2PCapsIndexGPU peer-to-peer capabilities index.
class
cuda_core.cuda.core.system.typing.GpuP2PStatusGPU peer-to-peer status.
class
cuda_core.cuda.core.system.typing.InforomObjectInfoROM objects types.
class
cuda_core.cuda.core.system.typing.SystemEventTypeSystem event types.
class
cuda_core.cuda.core.system.typing.TemperatureThresholdsTemperature threshold types.
class
cuda_core.cuda.core.system.typing.ThermalControllerThermal controller types.
class
cuda_core.cuda.core.system.typing.ThermalTargetThermal sensor targets.
class
cuda_core.cuda.core.typing.AddressModeTypeBoundary behavior for out-of-range texture coordinates.
class
cuda_core.cuda.core.typing.GraphMemoryTypeMemory space for a graph memory-allocation or free node.
class
cuda_core.cuda.core.typing.ReadModeTypeHow sampled values are returned to the kernel.
class
cuda_core.cuda.core.typing.SourceCodeTypeSource language passed to :class:`~cuda.core.Program`.
class
cuda_core.cuda.core.typing.VirtualMemoryAccessTypeAccess permissions for a virtual memory mapping.
class
cuda_core.cuda.core.utils._program_cache._abc.ProgramCacheResourceAbstract base class for compiled-program caches.
method
cuda_core.cuda.core.utils._program_cache._abc.ProgramCacheResource.clear() -> NoneRemove every entry from the cache.
method
cuda_core.cuda.core.utils._program_cache._abc.ProgramCacheResource.close() -> NoneRelease backend resources.
method
cuda_core.cuda.core.utils._program_cache._abc.ProgramCacheResource.get(key:bytes | str, default:bytes | None=None) -> bytes | NoneReturn ``self[key]`` or ``default`` if absent.
class
cuda_core.cuda.core.utils._program_cache._in_memory.InMemoryProgramCacheIn-memory program cache with LRU eviction.
func
cuda_pathfinder.cuda.pathfinder._binaries.find_nvidia_binary_utility.find_nvidia_binary_utility(utility_name:str) -> str | NoneLocate a CUDA binary utility executable.
func
cuda_pathfinder.cuda.pathfinder._dynamic_libs.load_dl_windows.ctypes_handle_to_unsigned_int(handle:ctypes.wintypes.HMODULE) -> intConvert ctypes HMODULE to unsigned int.
func
cuda_pathfinder.cuda.pathfinder._dynamic_libs.load_nvidia_dynamic_lib.load_nvidia_dynamic_lib(libname:str) -> LoadedDLLoad an NVIDIA dynamic library by name.
class
cuda_pathfinder.cuda.pathfinder._dynamic_libs.search_steps.FindResultA library file located on disk (not yet loaded).
func
cuda_pathfinder.cuda.pathfinder._dynamic_libs.search_steps.find_in_conda(ctx:SearchContext) -> FindResult | NoneSearch ``$CONDA_PREFIX``.
func
cuda_pathfinder.cuda.pathfinder._dynamic_libs.search_steps.find_in_cuda_path(ctx:SearchContext) -> FindResult | NoneSearch ``$CUDA_PATH`` / ``$CUDA_HOME``.
func
cuda_pathfinder.cuda.pathfinder._dynamic_libs.search_steps.find_in_site_packages(ctx:SearchContext) -> FindResult | NoneSearch pip wheel install locations.
func
cuda_pathfinder.cuda.pathfinder._headers.find_nvidia_headers.find_in_conda(desc:HeaderDescriptor) -> LocatedHeaderDir | NoneSearch ``$CONDA_PREFIX``.
func
cuda_pathfinder.cuda.pathfinder._headers.find_nvidia_headers.find_in_cuda_path(desc:HeaderDescriptor) -> LocatedHeaderDir | NoneSearch ``$CUDA_PATH`` / ``$CUDA_HOME``.
func
cuda_pathfinder.cuda.pathfinder._headers.find_nvidia_headers.find_in_site_packages(desc:HeaderDescriptor) -> LocatedHeaderDir | NoneSearch pip wheel install locations.
class
cuda_pathfinder.cuda.pathfinder._static_libs.find_bitcode_lib.LocatedBitcodeLibInformation about a located bitcode library.
func
cuda_pathfinder.cuda.pathfinder._static_libs.find_bitcode_lib.find_bitcode_lib(name:str) -> strFind the absolute path to a bitcode library.
func
cuda_pathfinder.cuda.pathfinder._static_libs.find_bitcode_lib.locate_bitcode_lib(name:str) -> LocatedBitcodeLibLocate a bitcode library by name.
class
cuda_pathfinder.cuda.pathfinder._static_libs.find_static_lib.LocatedStaticLibInformation about a located static library.
func
cuda_pathfinder.cuda.pathfinder._static_libs.find_static_lib.find_static_lib(name:str) -> strFind the absolute path to a static library.
func
cuda_pathfinder.cuda.pathfinder._static_libs.find_static_lib.locate_static_lib(name:str) -> LocatedStaticLibLocate a static library by name.
func
cuda_pathfinder.cuda.pathfinder._utils.env_vars.get_cuda_path_or_home() -> str | NoneGet CUDA Toolkit path from environment variables.
func
toolshed.check_cython_abi.parse_pxd_structs(pxd_path:Path) -> dictParse struct and cdef class field definitions from a .pxd file.
func
toolshed.find_skipped_tests.is_test_config_job(job_name:str) -> boolTrue if the job appears to be a wheel-based test configuration.
この情報について
掲載しているシグネチャは NVIDIA/cuda-python の公開ソースコードを
Python の ast モジュールで静的解析し、引数名・デフォルト値・
型注釈・戻り値型をそのまま抽出したものです。実装コードは保存していません。
詳しくは仕組みの解説をご覧ください。