Bot releases are hidden (Show)
Published by mitmul over 6 years ago
This is the release note of v2.4.0. See here for the complete list of solved issues and merged PRs.
rand
/randn
with examples (#911)Published by kmaehashi over 6 years ago
This is the release of v4.0.0b3. See here for the complete list of solved issues and merged PRs.
cupy.random.shuffle
performance has been improved. See #603 for details.shuffle
(#603, thanks @anaruse!)sum
for cupy.sparse
(#712)atleast_nd
(#819)size
function (#827)eliminate_zeros
for csc
matrix (#831)rand
for cupy.sparse
(#836)coo
initializer for SciPy sparse matrix (#857)None
in set_allocator
and set_pinned_memory_allocator
(#885)repeat
that behaved differently from NumPy (#670)linalg.norm
for complex input (#781, thanks @kohr-h!)ctxGetCurrent
(#837)out
argument in fusion (#868)cupy.size
overflow in 32-bit (#883)CArray
and CIndexer
(#892)testing/parameterized.py
from Chainer (#784)Published by bkvogel over 6 years ago
This is the release of v2.3.0. See here for the complete list of solved issues and merged PRs.
Published by delta2323 almost 7 years ago
This is the release of v4.0.0b2. See here for the complete list of solved issues and merged PRs.
cupy.moveaxis
(#684, thanks @fukatani!)default_casting
option to ufuncs (#720)None
as an argument for clip method (#802)matmul
to raise ValueError
on invalid shapes (#737)RandomState.choice
reproducibility (#741)stack
(#749)order=None
work like default order (#764, thanks @kohr-h!)order=None
of unravel_index
(#791, thanks @Hakuyume!)CUPY_SEED
environment variable to uint64
(#805, thanks @toslunar!)argmin
and argmax
(#774)numpy.stack
in numpy 1.9 (#803)numpy.matmul
in numpy 1.9 (#817)ElementwiseKernel.__call__
(#786)accept_error
argument in testing.numpy_cupy_raises
(#787)testing.numpy_cupy_equal
(#804)Published by beam2d almost 7 years ago
This is the release of v2.2.0. See here for the complete list of solved issues and merged PRs.
RandomState.choice
reproducibility (#775)stack
function bug (#798)matmul
to raise ValueError
on invalid shapes (#816)CUPY_SEED
environment variable to uint64
(#822, thanks @toslunar!)dtype
option of randint
which is introduced in NumPy v1.11 (#830)ElementwiseKernel.__call__
docs (#810)Published by beam2d almost 7 years ago
This is a hotfix of v2.1. It contains a fix of the issue #766 that cupy.random
functions raises an error when either CUPY_SEED
or CHAINER_SEED
is set. This issue has been fixed via #805, which is cherry-picked for this hot-fix.
This release does not contain any other updates from v2.1.0.
Published by hvy almost 7 years ago
This is the release of v2.1.0. See here for the complete list of solved issues and merged PRs.
argpartition
(#608)blackman
, hamming
, hanning
sparse.coo_matrix
initialization with other types of sparse matrices (#626)dtype
argument in random.randint
(#706)sparse.csc_matrix.__mul__
(#625)random.RandomState.interval
(#633)random.RandomState.seed
to only accept integer types (#709)testing.fix_random
(#648)var
and std
to correctly handle ddof
argument (#711, thanks @stevendbrown!)testing.numpy_cupy_raises
(#637, thanks @Hakuyume!)linalg
(#651)CUPY_TEST_GPU_LIMIT
environment variable (#677)ComplexWarning
in numpy.pad
for NumPy 1.11 or older (#690)where
to use different seeds for different arrays (#710)test_einsum
(#740)Published by kmaehashi almost 7 years ago
This is the release of v4.0.0b1. See here for the complete list of solved issues and merged PRs.
As the version number indicates, we decided to name the next major version of CuPy v4 instead of v3 to align the versioning with Chainer.
From this version, you can install compatible versions of Chainer and CuPy by specifying the same version number for both.
cupy.fft
(#477)
fft
, ifft
, fft2
, ifft2
, fftn
, ifftn
rfft
, irfft
, rfft2
, irfft2.
, rfftn
, irfftn
hfft
, ihfft
fftfreq
, rfftfreq
, fftshift
, ifftshift
random.RandomState.tomaxint
(#389)sparse.csr_matrix.eliminate_zeros
and sparse.coo_matrix.eliminate_zeros
(#398)linalg.tensorinv
(#464)unravel_index
(#632, thanks @Hakuyume!)percentile
(#643)random.set_random_state
(#704)einsum
(#410, thanks @fukatani!)dtype
argument in random.randint
(#567)sparse.coo_matrix
initialization with other types of sparse matrices (#573)testing.fix_random
(#640)var
and std
to correctly handle ddof
argument (#693, thanks @stevendbrown!)random.RandomState.tomaxint
for Windows (#658)concatenate
by using continuous copies (#452, thanks @uchida!)sparse.csc_matrix.__mul__
(#572)linear_launch
(#673)random.RandomState.interval
(#583)random.RandomState.seed
to only accept integer types (#688)testing.numpy_cupy_raises
(#634, thanks @Hakuyume!)linalg
(#650)CUPY_TEST_GPU_LIMIT
environment variable (#662)test_einsum
(#679)ComplexWarning
in numpy.pad
for NumPy 1.11 or older (#689)where
to use different seeds for different arrays (#703)test_einsum
(#726)test_fft
for NumPy 1.9 or older (#727)Published by bkvogel about 7 years ago
This is a major release of CuPy v2.0.0. All of the updates since the previous major version (v1.0.0) can be found in the release notes below:
cupy.sparse
is a module that implements scipy.sparse
API using CUDA and cuSPARSE. We now have basic features for using sparse matrices on GPU.cupy.nonzero
for corner cases (#504)PooledMemory
(#507)PinnedMemory
(#510)RandomState.choice
(#560)broadcast
for corner cases (#577)csrmm2
to support transa (#601)csrmv
(#607)get_array_module
to be aware of spmatrix
(#586)RandomState.interval
(#585)random.normal
double memory consumption (#592)memory_hooks
(#506)cupy.all
and cupy.any
function (#514)mock.patch
instead of directly replacing function with Mock
(#610)print()
in tests (#509)csrgemm
(#602)Published by gwtnb about 7 years ago
This is the release of CuPy v3.0.0a1. See here for the complete list of solved issues and merged PRs.
cudnnGetTensor4dDescriptor
for fp16 BatchNormalization support in Chainer (#492, thanks @anaruse!)cupy.sparse.random
(#557)cupy.argpartition
(#294)PooledMemory
(#480)PinnedMemory
(#481)cupy.nonzero
for corner cases (#498)broadcast
for corner cases (#543)broadcast_arrays
return type (#545)RandomState.choice
(#556)csrmm2
to support transa (#565)csrmv
(#571)dtype
option in numpy.random.randint
which is introduced in NumPy v1.11 (#574)get_array_module
to be aware of spmatrix
(#568)vector
to improve free memory searching in malloc
(#476)RandomState.interval
(#559)random.normal
double memory consumption (#562)memory_hooks
(#502)cupy.all
and cupy.any
function (#511)Published by mitmul about 7 years ago
This release includes bug fixes and improvements to the documentation and tests. See the list for the complete list of solved issues and merged PRs.
UnicodeDecodeError
. (#378, #379)__dealloc__
and use __del__
instead. (#411)ndarray.view
when the itemsize
of the dtype
changes. (#416)ndarray.diagonal
between NumPy and CuPy. (#436)strides
argument from docstring. (#366)matmul
. (#412)str
literals. (#429)randint
instead of random_integer
that is deprecated. (#430)testing.assert_warns
and test deprecation warning of Memory.free_all_free
. (#431)RandomState
when NumPy < 1.11.0. (#438)Published by bkvogel about 7 years ago
This is the release of CuPy v2.0.0rc1. See here for the complete list of solved issues and merged PRs.
order
argument of copy
from ’C’
to ’K’
(#159)order
and subok
arguments to array
(#167). It breaks the compatibility of positional arguments.angle
, conj
, imag
, real
(#232)einsum
(#199, thanks @fukatani!)linalg.solve
(#207), linalg.tensorsolve
(#215), linalg.inv
(#441), linalg.pinv
(#459)random.shuffle
(#216, thanks @KotaroSetoyama!)partition
(#270)dia_matrix
(#313, #321, #320, #450)eye
(#399), spdiags
(#388) and identity
(#358)csr_matrix
and csc_matrix
are improved: __mul__
(#239), __rmul__
(#300), __getitem__
(#240, #301, #302), dot
(#351, #352)csr_matrix
, csc_matrix
, and coo_matrix
support shape
argument (#316, #375)order
argument in toarray
method of csc and csr (#311)__pow__
(#359)scipy.sparse
matrix to cupy.sparse
(#370)argsort
for arrays of rank two or more (#288)replace=False
in random.choice
(#453)sync
option to time_range
(#474, thanks @anaruse!)coo_matrix
(#328)spmatrix
(#356)__dealloc__()
, use __del__()
instead. (#381)pow
test (#421)randint
instead of random_integer
, which is deprecated (#425)diagonal
(#428, thanks @fukatani!)six.assertRegex
(#432)cudaErrorMemoryAllocation
occurred (#314)setup.py
develop to build faster (#309)cupy_thrust.cu
(#369)_tril()
and _triu()
with an ElementwiseKernel
(#377)ElementwiseKernel
(#391)cumsum
(#414)AxisError
to maintain compatibility to multiple versions of NumPy (#437)tensordot_core
(#465)flip
(#468)None
instead of set()
to improve memory allocation performance (#475)setup.py develop
to build faster (#309)tuple_less
(#368)cupy.sparse
reference (#299, #303)README.md
(#334)strides
argument from docstring (#361)matmul
arguments (#384, thanks @hvy!)linalg.einsum
(#405)sparse.spdiags
docstring. (#426)linalg
(#456)sparse.issparse
(#470)cuda.cusolver_enabled
flag (#374)A
property and its test (#407)cumsum
test (#414)transpose
when axes is not None
(#420)pow
test (#421)asfptype
(#423)assert_warns
(#424)randint
instead of random_integer
that is deprecated (#425)six.assertRegex
(#432)__iter__
of csr_matrix
(#449)tocsc
behavior for an empty dia
matrix (#451)tensorsolve
(#454)clip
tests in Windows (#467)Published by niboshi about 7 years ago
This is a minor release. See https://github.com/cupy/cupy/milestone/8?closed=1 for the complete list of solved issues and merged PRs.
cupy.sparse
is a module that implements scipy.sparse
API using CUDA and cuSPARSE. We now have basic features for using sparse matrices on GPU.
__add__
, __radd__
, __sub__
and __rsub__
for CSR and CSC (#238)toarray
in cupy.sparse.spmatrix
(#312)NotImplemented
instead of NotImplementedError
(#330)csc2dense
to convert csr-matrix to dense (#305)We are planning to add more features to cupy.sparse
in upcoming releases.
The memory pool implementation is greatly updated. It is based on best-fit allocation with coalescing. When there are a large number of allocations with different sizes (e.g. NLP applications), the memory usage is improved and the number of re-allocations is reduced (which also reduces the running time).
For example, the memory usage of the sequence-to-sequence code using Chainer (chainer/chainer#2070) is reduced from 12GiB (which means the process is using all of the available GPU memory) to 3GiB, and the number of memory reallocations from 20 times to 0 times.
It may increase the memory usage in some cases, although the amount of additional usage is small in practice (see the benchmark results in #168).
You can use this memory allocator by calling cupy.cuda.set_allocator(cupy.cuda.MemoryPool().malloc)
(when using Chainer, it is called by default).
cupy.linalg.det
(#96)cupy.sort
to sort arrays along arbitrary axis (#229)RangeStart
and RangeEnd
for NVIDIA visual profiler (nvvp) (#246)cupy.is_available()
which takes account of device availability (#247)cupy.msort
(#251, #329)cupy.copyto
function to treat multiple GPUs correctly (#220)deepcopy
with multiple devices (#254)cupy.argsort
for non-contiguous arrays (#284)ldexp
on Windows (#278)cupy.argsort
performance (#285)cupy.cuda.thrust_enabled
to check Thrust enabled (#224)cupy.testing.for_all_dtypes
(#269)cupy.argsort
tests (#282)cupy.argsort
(#223)Published by bkvogel about 7 years ago
This release includes bug fixes and improvements to the documentation and tests. See the list for the complete list of solved issues and merged PRs.
allocation_unit_size
from 256 to 512 (#256)setup.py
develop (#293)cupy.array
(#259)VisibleDeprecationWarning
in indexing tests (#261)Published by delta2323 over 7 years ago
This release includes bug fixes and improvements on documents and tests. See the list for the complete list of solved issues and merged PRs.
cupy.random.get_random_state
(#77, #99)Published by unnonouno over 7 years ago
This is the release of CuPy v2.0.0a1. See here for the complete list of solved issues and merged PRs.
cupy.msort
(#150)cupy.lexsort
(#132)cupy.argsort
(#67)cupy.sort
sorting arrays with two or more rank along last axis (#186, #187)cupy.sort
support arrays with rank two or more. (#152)cupy.linalg.slogdet
(#95)cupy.linalg.matrix_rank
(#97)cupy.linalg.eigh
and cupy.linalg.eigvalsh
(#46)cupy.sparse.spmatrix
, a base class of sparse matrices (#40)cupy.mgrid
and cupy.ogrid
(#145, thanks @iory!)cupy.random.multinomial
(#85)cupy.cumprod
(#110, thanks @ronekko!)total_bytes()
, free_bytes()
, and used_bytes()
methods to memory pool (#184)order
option in astype
(#111) and copy
(#112)cupy.fuse
now does not require parentheses (#43)ndim
to CArray
and CIndexer
(#160, #161)cupy.array
for 0-dim values (#157)cupy.count_nonzero
return an array instead of int
to avoid device-to-host synchronization (#154)assert_array_list_equal
(#205)cupy.atleast_nd
functions (#142)out
argument in fusion (#209, #213)cupy.array
on multiple GPU environment (#122, #135)copy
argument of ndarray.astype
(#118, #121)cupy.random.get_random_state
(#77, #78)cupy.cumsum
(#90, thanks @ronekko!)__getitem__
and __setitem__
for ndarray (#89, thanks @yuyu2172!)cupy.random.choice
test (#98, #104)VisibleDeprecationWarning
in indexing tests (#202)RandomState.interval
test (#175)Published by bkvogel over 7 years ago
This is the release of CuPy v1.
This release also contains updates of CuPy included in Chainer v1.23.0 and v1.24.0. See the release note of Chainer v1.23.0 and the release note of Chainer v1.24.0 for the details.
The set of supported versions of CUDA and cuDNN is changed from Chainer v1.x as follows.
Note: We had originally planned to include NVRTC support for the just-in-time compilation of kernels via pynvrtc
, but we found that there is no guarantee on pynvrtc
being compatible with old versions of CUDA, so we decided to make our own wrapper instead. Unfortunately, it cannot be included in this version. We are planning to add NVRTC support in the next version.
cupy.sort
function (#55, #66, #68)CUPY_SEED
enviroment variable (#44)cupy.copyto
from Python scalar (#38)cupy.random.choise
(#84)Published by beam2d over 7 years ago
This is the beta release of CuPy v1.
This release only contains updates of Chainer v1.22.0 and minor updates of documentation and installation. See the release note of Chainer v1.22.0 for the details.
Published by beam2d over 7 years ago
This is the first alpha of CuPy v1!
At the moment, the API and the implementation is equivalent to CuPy included in Chainer v1.21.0, while the installation step is slightly different (you need to install cupy package instead of chainer). See our official documentation for the installation procedure.
Note that the development is currently runinng at pfnet/chainer repository. We will work on adding a new feature only included in this independent CuPy code base. We will also catch up with the updates on pfnet/chainer repository, and so feel free to send any issues and patches for CuPy to pfnet/chainer repository.