API reference

Generated from the docstrings. Each name has one import path, through the module it is listed under (xp.kernels.CudaKernel, xp.rng.random_streams). For an overview of the module layout and the NumPy-like namespace, see the API overview.

• Top level: cunumpy: The NumPy/CuPy namespace (see the API overview) plus backend selection, array inspection and conversion, and host execution helpers.
• Kernels: cunumpy.kernels: Kernels that run on either backend: dispatch, host implementations, CUDA and Metal kernels, fusion.
• Kernel arguments: cunumpy.arguments: Structs and argument objects passed to CUDA kernels.
• CUDA runtime: cunumpy.cuda: Devices, memory, streams, events, debug mode and CUDA source tools. These do nothing on the NumPy backend.
• Random numbers: cunumpy.rng: Reproducible random streams and counter-based Philox generators.
• Algorithms: cunumpy.algorithms: Morton keys, sorting by key, segment sums and compaction.
• Optimization: cunumpy.optimize: Batched Newton and secant iterations on the device, and SciPy’s fsolve, root and minimize with callbacks on the device.
• Integration: cunumpy.integrate: SciPy’s quad and odeint with callbacks on the device.
• MPI: cunumpy.mpi: MPI buffers, CUDA-aware MPI detection and synchronization. cunumpy.mpi also re-exports the MPI stand-ins of maybempi, documented there: get_mpi, is_serial, launched_under_mpi, local_rank, OVERRIDE_VARIABLE, SerialMPI, SerialComm, SerialRequest and SerialStatus.
• Profiling: cunumpy.profiling: Timed regions, NVTX ranges and host/device transfer counting.
• Memory: cunumpy.memory: Pinned host staging, asynchronous host copies and device mirrors.
• PETSc: cunumpy.petsc: PETSc vectors that share memory with arrays.
• Kernel testing: cunumpy.kernel_testing: Pytest helpers for host/CUDA kernel pairs. Not imported by import cunumpy.