CuPy

CuPy implements NumPy/SciPy-compatible APIs, as well as features to write user-defined GPU kernels or access low-level APIs.{{cite web

| url = https://docs.cupy.dev/en/latest/overview.html

| title = Overview

| work = CuPy documentation

| access-date = 18 June 2022}}{{cite web

| url = https://docs.cupy.dev/en/latest/reference/comparison.html

| title = Comparison Table

| work = CuPy documentation

| access-date = 18 June 2022

}}

The same set of APIs defined in the NumPy package ({{code|numpy.*}}) are available under {{code|cupy.*}} package.

• Multi-dimensional array ({{code|cupy.ndarray}}) for boolean, integer, float, and complex data types
• Module-level functions
• Linear algebra functions
• Fast Fourier transform
• Random number generator

The same set of APIs defined in the SciPy package ({{code|scipy.*}}) are available under {{code|cupyx.scipy.*}} package.

• Sparse matrices ({{code|cupyx.scipy.sparse.*_matrix}}) of CSR, COO, CSC, and DIA format
• Discrete Fourier transform
• Advanced linear algebra
• Multidimensional image processing
• Sparse linear algebra
• Special functions
• Signal processing
• Statistical functions

• Kernel templates for element-wise and reduction operations
• Raw kernel (CUDA C/C++)
• Just-in-time transpiler (JIT)
• Kernel fusion

• Distributed communication package ({{code|cupyx.distributed}}), providing collective and peer-to-peer primitives

• Stream and event
• Memory pool
• Profiler
• Host API binding
• CUDA Python support{{cite web

|url = https://developer.nvidia.com/cuda-python

|title = CUDA Python {{!}} NVIDIA Developer

|access-date = 21 June 2022

}}

• DLPack{{cite web

| url = https://dmlc.github.io/dlpack/latest/

| title = Welcome to DLPack's documentation!

| work = DLPack 0.6.0 documentation

| access-date = 21 June 2022

}}

• CUDA Array Interface{{Cite web

| title = CUDA Array Interface (Version 3)

| work = Numba 0.55.2+0.g2298ad618.dirty-py3.7-linux-x86_64.egg documentation

| access-date = 21 June 2022

| url = https://numba.readthedocs.io/en/stable/cuda/cuda_array_interface.html

}}

• NEP 13 ({{code|__array_ufunc__}}){{Cite web

| title = NEP 13 — A mechanism for overriding Ufuncs — NumPy Enhancement Proposals

| work = numpy.org

| access-date = 21 June 2022

| url = https://numpy.org/neps/nep-0013-ufunc-overrides.html

}}

• NEP 18 ({{code|__array_function__}}){{Cite web

| title = NEP 18 — A dispatch mechanism for NumPy's high level array functions — NumPy Enhancement Proposals

| work = numpy.org

| access-date = 21 June 2022

| url = https://numpy.org/neps/nep-0018-array-function-protocol.html

}}{{cite Q|Q99413970|display-authors=3}}

• Array API Standard{{Cite web

| title = 2021 report - Python Data APIs Consortium

| access-date = 21 June 2022

| url = https://data-apis.org/files/2021_annual_report_DataAPIs_Consortium.pdf

}}{{Cite web

| title = Purpose and scope

| work = Python array API standard 2021.12 documentation

| access-date = 21 June 2022

| url = https://data-apis.org/array-api/latest/purpose_and_scope.html

}}

>>> import cupy as cp

>>> x = cp.array([1, 2, 3])

>>> x

array([1, 2, 3])

>>> y = cp.arange(10)

>>> y

array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])

>>> import cupy as cp

>>> x = cp.arange(12).reshape(3, 4).astype(cp.float32)

>>> x

array([[ 0., 1., 2., 3.],

[ 4., 5., 6., 7.],

[ 8., 9., 10., 11.]], dtype=float32)

>>> x.sum(axis=1)

array([ 6., 22., 38.], dtype=float32)

>>> import cupy as cp

>>> kern = cp.RawKernel(r'''

... extern "C" __global__

... void multiply_elemwise(const float* in1, const float* in2, float* out) {

... int tid = blockDim.x * blockIdx.x + threadIdx.x;

... out[tid] = in1[tid] * in2[tid];

... }

... ''', 'multiply_elemwise')

>>> in1 = cp.arange(16, dtype=cp.float32).reshape(4, 4)

>>> in2 = cp.arange(16, dtype=cp.float32).reshape(4, 4)

>>> out = cp.zeros((4, 4), dtype=cp.float32)

>>> kern((4,), (4,), (in1, in2, out)) # grid, block and arguments

>>> out

array([[ 0., 1., 4., 9.],

[ 16., 25., 36., 49.],

[ 64., 81., 100., 121.],

[144., 169., 196., 225.]], dtype=float32)

• spaCy{{Cite web

| title = Install spaCy

| work = spaCy Usage Documentation

| url = https://spacy.io/usage#gpu

| access-date = 21 June 2022

}}{{cite book

| last1 = Patel

| first1 = Ankur A.

| last2 = Arasanipalai

| first2 = Ajay Uppili

| title = Applied Natural Language Processing in the Enterprise

| publisher = O'Reilly Media, Inc.

| edition = 1st

| date = May 2021

| isbn = 9781492062578

| page = 68

}}

• XGBoost{{Cite web

| title = Python Package Introduction

| work = xgboost 1.6.1 documentation

| access-date = 21 June 2022

| url = https://xgboost.readthedocs.io/en/stable/python/python_intro.html#data-interface

}}

• {{not a typo|turboSETI}} (Berkeley SETI){{Cite web

| title = UCBerkeleySETI/turbo_seti: turboSETI -- python based SETI search algorithm.

| website = GitHub

| url = https://github.com/UCBerkeleySETI/turbo_seti#turbo_seti

| access-date = 21 June 2022

}}

• NVIDIA RAPIDS{{Cite web

| title = Open GPU Data Science {{!}} RAPIDS

| access-date = 21 June 2022

| url = https://rapids.ai/

}}{{Cite web

| title = API Docs

| work = RAPIDS Docs

| access-date = 21 June 2022

| url = https://docs.rapids.ai/api

}}{{Cite web

| title = Efficient Data Sharing between CuPy and RAPIDS

| access-date = 21 June 2022

| url = https://medium.com/rapids-ai/using-rapids-memory-manager-with-cupy-8d08fe8f58fa

}}{{Cite web

| title = 10 Minutes to cuDF and CuPy

| access-date = 21 June 2022

| url = https://medium.com/rapids-ai/10-minutes-to-cudf-and-cupy-e131cac0439b

}}

• {{not a typo|einops}}

{{cite conference

| url = https://openreview.net/forum?id=oapKSVM2bcj

| title = Einops: Clear and Reliable Tensor Manipulations with Einstein-like Notation

| last1 = Alex

| first1 = Rogozhnikov

| date = 2022

| publisher = International Conference on Learning Representations

}}{{Cite web

| title = arogozhnikov/einops: Deep learning operations reinvented (for pytorch, tensorflow, jax and others)

| website = GitHub

| url = https://github.com/arogozhnikov/einops

| access-date = 21 June 2022

}}

• scikit-learn{{Cite web

| title = Array API support (experimental) — scikit-learn documentation

| url = https://scikit-learn.org/stable/modules/array_api.html

| access-date = 8 September 2024

}}

• MONAI
• Chainer

{{cite conference

| url = https://dl.acm.org/doi/10.1145/3292500.3330756

| title = Chainer: A Deep Learning Framework for Accelerating the Research Cycle

| last1 = Tokui

| first1 = Seiya

| last2 = Okuta

| first2 = Ryosuke

| last3 = Akiba

| first3 = Takuya

| last4 = Niitani

| first4 = Yusuke

| last5 = Ogawa

| first5 = Toru

| last6 = Saito

| first6 = Shunta

| last7 = Suzuki

| first7 = Shuji

| last8 = Uenishi

| first8 = Kota

| last9 = Vogel

| first9 = Brian

| last10 = Vincent

| first10 = Hiroyuki Yamazaki

| date = 2019

| publisher = Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining

| doi = 10.1145/3292500.3330756

| url-access = subscription

}}

{{Portal|Free software}}

• Array programming
• List of numerical-analysis software
• Dask

{{reflist}}

• {{Official website|http://cupy.dev/}}
• {{GitHub|cupy/cupy}}

{{SciPy ecosystem}}

{{DEFAULTSORT:CuPy}}

Category:Array programming languages

Category:Articles with example Python (programming language) code

Category:Free mathematics software

Category:Free science software

Category:Numerical analysis software for Linux

Category:Numerical analysis software for Windows

Category:Numerical programming languages

Category:Python (programming language) scientific libraries

Category:Software using the MIT license