sdkagent

cuda-python API reference

53 public APIs from cuda-python (NVIDIA/cuda-python) — 27 classes, 18 functions, 8 methods. Signatures extracted by static analysis of the actual source.

Repository: NVIDIA/cuda-python

KindCount
Classes27
Functions18
Methods8

API list

funccuda_bindings.cuda.bindings.utils._ptx_utils.get_ptx_ver(ptx:str) -> str
Extract the PTX ISA version string from PTX source code.
classcuda_core.cuda.core._host.Host
Host (CPU) location for managed-memory operations.
funccuda_core.cuda.core._utils.enum_explanations_helpers.clean_enum_member_docstring(doc:str | None) -> str | None
Turn an enum member ``__doc__`` into plain text.
methodcuda_core.cuda.core.checkpoint.Process.lock(timeout_ms:int=0) -> None
Lock this process, blocking further CUDA API calls.
methodcuda_core.cuda.core.checkpoint.Process.pid() -> int
Process ID of the CUDA process.
methodcuda_core.cuda.core.checkpoint.Process.restore(gpu_mapping:Mapping[Any, Any] | None=None) -> None
Restore this checkpointed process.
methodcuda_core.cuda.core.checkpoint.Process.restore_thread_id() -> int
CUDA restore thread ID for this process.
methodcuda_core.cuda.core.checkpoint.Process.state() -> _ProcessStateType
CUDA checkpoint state for this process.
classcuda_core.cuda.core.system.typing.AddressingMode
Addressing mode of a device.
classcuda_core.cuda.core.system.typing.AffinityScope
Scope for affinity queries.
classcuda_core.cuda.core.system.typing.ClockId
Clock Ids.
classcuda_core.cuda.core.system.typing.ClockType
Clock types.
classcuda_core.cuda.core.system.typing.ClocksEventReasons
Reasons for a clocks event.
classcuda_core.cuda.core.system.typing.CoolerControl
Cooler control type.
classcuda_core.cuda.core.system.typing.CoolerTarget
Cooler target.
classcuda_core.cuda.core.system.typing.DeviceArch
Device architecture.
classcuda_core.cuda.core.system.typing.FanControlPolicy
Fan control policies.
classcuda_core.cuda.core.system.typing.GpuP2PCapsIndex
GPU peer-to-peer capabilities index.
classcuda_core.cuda.core.system.typing.GpuP2PStatus
GPU peer-to-peer status.
classcuda_core.cuda.core.system.typing.InforomObject
InfoROM objects types.
classcuda_core.cuda.core.system.typing.SystemEventType
System event types.
classcuda_core.cuda.core.system.typing.TemperatureThresholds
Temperature threshold types.
classcuda_core.cuda.core.system.typing.ThermalController
Thermal controller types.
classcuda_core.cuda.core.system.typing.ThermalTarget
Thermal sensor targets.
classcuda_core.cuda.core.typing.AddressModeType
Boundary behavior for out-of-range texture coordinates.
classcuda_core.cuda.core.typing.GraphMemoryType
Memory space for a graph memory-allocation or free node.
classcuda_core.cuda.core.typing.ReadModeType
How sampled values are returned to the kernel.
classcuda_core.cuda.core.typing.SourceCodeType
Source language passed to :class:`~cuda.core.Program`.
classcuda_core.cuda.core.typing.VirtualMemoryAccessType
Access permissions for a virtual memory mapping.
classcuda_core.cuda.core.utils._program_cache._abc.ProgramCacheResource
Abstract base class for compiled-program caches.
methodcuda_core.cuda.core.utils._program_cache._abc.ProgramCacheResource.clear() -> None
Remove every entry from the cache.
methodcuda_core.cuda.core.utils._program_cache._abc.ProgramCacheResource.close() -> None
Release backend resources.
methodcuda_core.cuda.core.utils._program_cache._abc.ProgramCacheResource.get(key:bytes | str, default:bytes | None=None) -> bytes | None
Return ``self[key]`` or ``default`` if absent.
classcuda_core.cuda.core.utils._program_cache._in_memory.InMemoryProgramCache
In-memory program cache with LRU eviction.
funccuda_pathfinder.cuda.pathfinder._binaries.find_nvidia_binary_utility.find_nvidia_binary_utility(utility_name:str) -> str | None
Locate a CUDA binary utility executable.
funccuda_pathfinder.cuda.pathfinder._dynamic_libs.load_dl_windows.ctypes_handle_to_unsigned_int(handle:ctypes.wintypes.HMODULE) -> int
Convert ctypes HMODULE to unsigned int.
funccuda_pathfinder.cuda.pathfinder._dynamic_libs.load_nvidia_dynamic_lib.load_nvidia_dynamic_lib(libname:str) -> LoadedDL
Load an NVIDIA dynamic library by name.
classcuda_pathfinder.cuda.pathfinder._dynamic_libs.search_steps.FindResult
A library file located on disk (not yet loaded).
funccuda_pathfinder.cuda.pathfinder._dynamic_libs.search_steps.find_in_conda(ctx:SearchContext) -> FindResult | None
Search ``$CONDA_PREFIX``.
funccuda_pathfinder.cuda.pathfinder._dynamic_libs.search_steps.find_in_cuda_path(ctx:SearchContext) -> FindResult | None
Search ``$CUDA_PATH`` / ``$CUDA_HOME``.
funccuda_pathfinder.cuda.pathfinder._dynamic_libs.search_steps.find_in_site_packages(ctx:SearchContext) -> FindResult | None
Search pip wheel install locations.
funccuda_pathfinder.cuda.pathfinder._headers.find_nvidia_headers.find_in_conda(desc:HeaderDescriptor) -> LocatedHeaderDir | None
Search ``$CONDA_PREFIX``.
funccuda_pathfinder.cuda.pathfinder._headers.find_nvidia_headers.find_in_cuda_path(desc:HeaderDescriptor) -> LocatedHeaderDir | None
Search ``$CUDA_PATH`` / ``$CUDA_HOME``.
funccuda_pathfinder.cuda.pathfinder._headers.find_nvidia_headers.find_in_site_packages(desc:HeaderDescriptor) -> LocatedHeaderDir | None
Search pip wheel install locations.
classcuda_pathfinder.cuda.pathfinder._static_libs.find_bitcode_lib.LocatedBitcodeLib
Information about a located bitcode library.
funccuda_pathfinder.cuda.pathfinder._static_libs.find_bitcode_lib.find_bitcode_lib(name:str) -> str
Find the absolute path to a bitcode library.
funccuda_pathfinder.cuda.pathfinder._static_libs.find_bitcode_lib.locate_bitcode_lib(name:str) -> LocatedBitcodeLib
Locate a bitcode library by name.
classcuda_pathfinder.cuda.pathfinder._static_libs.find_static_lib.LocatedStaticLib
Information about a located static library.
funccuda_pathfinder.cuda.pathfinder._static_libs.find_static_lib.find_static_lib(name:str) -> str
Find the absolute path to a static library.
funccuda_pathfinder.cuda.pathfinder._static_libs.find_static_lib.locate_static_lib(name:str) -> LocatedStaticLib
Locate a static library by name.
funccuda_pathfinder.cuda.pathfinder._utils.env_vars.get_cuda_path_or_home() -> str | None
Get CUDA Toolkit path from environment variables.
functoolshed.check_cython_abi.parse_pxd_structs(pxd_path:Path) -> dict
Parse struct and cdef class field definitions from a .pxd file.
functoolshed.find_skipped_tests.is_test_config_job(job_name:str) -> bool
True if the job appears to be a wheel-based test configuration.

About this data

These signatures were extracted from the public source of NVIDIA/cuda-python using Python's ast module. Argument names, default values, type annotations and return types are taken verbatim from the code. Implementation bodies are never stored. See how it works for details.

Back to all 805 libraries