API Reference

• Index

• Module Index

---

• The N-dimensional array (ndarray)
  • cupy.ndarray
  • Conversion to/from NumPy arrays
  • Code compatibility features
• Universal functions (cupy.ufunc)
  • ufunc
  • Available ufuncs
  • Generalized Universal Functions
• Routines (NumPy)
  • Array creation routines
  • Array manipulation routines
  • Bit-wise operations
  • Data type routines
  • Discrete Fourier Transform (cupy.fft)
  • Functional programming
  • Indexing routines
  • Input and output
  • Linear algebra (cupy.linalg)
  • Logic functions
  • Mathematical functions
  • Miscellaneous routines
  • Polynomials
  • Random sampling (cupy.random)
  • Set routines
  • Sorting, searching, and counting
  • Statistics
  • Test support (cupy.testing)
  • Window functions
• Routines (SciPy)
  • Discrete Fourier transforms (cupyx.scipy.fft)
  • Legacy discrete fourier transforms (cupyx.scipy.fftpack)
  • Interpolation (cupyx.scipy.interpolate)
  • Linear algebra (cupyx.scipy.linalg)
  • Multidimensional image processing (cupyx.scipy.ndimage)
  • Signal processing (cupyx.scipy.signal)
  • Signal processing windows (cupyx.scipy.signal.windows)
  • Sparse matrices (cupyx.scipy.sparse)
  • Sparse linear algebra (cupyx.scipy.sparse.linalg)
  • Compressed sparse graph routines (cupyx.scipy.sparse.csgraph)
  • Spatial algorithms and data structures (cupyx.scipy.spatial)
  • Distance computations (cupyx.scipy.spatial.distance)
  • Special functions (cupyx.scipy.special)
  • Statistical functions (cupyx.scipy.stats)
• CuPy-specific functions
  • cupyx.rsqrt
  • cupyx.scatter_add
  • cupyx.scatter_max
  • cupyx.scatter_min
  • cupyx.empty_pinned
  • cupyx.empty_like_pinned
  • cupyx.zeros_pinned
  • cupyx.zeros_like_pinned
  • non-SciPy compat Signal API
  • Profiling utilities
  • DLPack utilities
  • Automatic Kernel Parameters Optimizations (cupyx.optimizing)
• Low-level CUDA support
  • Device management
  • Memory management
  • Memory hook
  • Streams and events
  • Graphs
  • Texture and surface memory
  • NVTX
  • NCCL
  • Version
  • Runtime API
• Custom kernels
  • cupy.ElementwiseKernel
  • cupy.ReductionKernel
  • cupy.RawKernel
  • cupy.RawModule
  • cupy.fuse
  • JIT kernel definition
  • Kernel binary memoization
• Distributed
  • Communication between processes
  • ndarray distributed across devices
• Environment variables
  • For runtime
  • For installation
• Comparison Table
  • NumPy / CuPy APIs
  • SciPy / CuPy APIs
• Python Array API Support