This is an automated email from the ASF dual-hosted git repository.
lausen pushed a change to branch leezu-patch-1
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.
discard 8a9599b Disable MKL_USE_STATIC_LIBS by default
add 2fff11d [OpPerf] Fix axis_shape and function mismatch for LTS (#17894)
add 4c0f763 Fix a typo (#17979)
add 1b107a0 Remove redundant condition in np_matrix_op.cc (#17933)
add 84b0ddd Add USE_DIST_KVSTORE=ON to GPU build (#17911)
add 03b8146 Skip test_kvstore_gpu.test_rsp_push_pull (#17983)
add ff234db Skip test_gluon_data.py on OSX (#17969)
add c244f9f [MXNET-#16795] Byteps-KVStore: Intergrate Byteps into mxnet
as new type of kvstore backend (#17555)
add 5adcbf8 GPU gemms true fp16 (#17466)
add c3c76a8 Optimize AddTakeGrad Tensor Sum (#17906)
add 002d4f1 * impl - FFi for linalg op (#17795)
add 79c576b [ONNX export] Fixing spatial export for batchnorm (#17711)
add 892f982 * impl - linalg.lstsq for cpu (#17950)
add a960f5a ffi_array_split, v/h/dsplit (#17873)
add f906a02 ffi_atleast_1/2/3d (#17897)
add 16ddc6d Custom Operator Random Number Generator Support (#17762)
add b7f7525 dnnl v1.2.2 (#17991)
add 13841dd [mkldnn] optimize for mkldnn batchnorm backward (#17902)
add da95add Fix vector access out of bound in MKLDNNConvolutionBackward
(#17997)
add 6cc990c Revert "[MXNET-#16795] Byteps-KVStore: Intergrate Byteps into
mxnet as new type of kvstore backend (#17555)" (#17998)
add d97dead [Numpy] allow mix integer dtypes for power/add/multiply
(#17921)
add 178b98e [Numpy] OP_interp (#17793)
add d1616c9 [numpy] FFI binary bitwise ops (#17812)
add 664889a [MKL-DNN] Integrate Conv3d and Pool3d/1d (#17884)
add 5ff2994 [Website 2.0] General Version Dropdown (#17948)
add 6a931c7 Add np.linalg.qr (#17851)
add 0eeb337 Improve activation backward (#17973)
add d8c7293 fix np flip when axis input contains negative number (#17880)
add 0d87aa8 [numpy] FFI for insert \ delete \ matmul etc. (#17759)
add 3dab617 add: numpy rollaxis (#17865)
add af76466 add: numpy op tril_indices (#17904)
add 8f82cd8 [MKLDNN] support using any format in pooling backward (#17900)
add 57c785a [Numpy][Bug Fix] Fix Wrong Result in Numpy Operator `where`
(#17899)
add e4afe50 * impl debug - FFI for linalg multioutput op (#17879)
add 2be2027 Add instructions on distributed MXNet with Horovod on
Kubernetes (#17974)
add 249b9a1 Fix cudnn Dropout reproducibility (#17547)
add a6fef3f [Website 2.0] Nightly Build for v2.x (master) (#17957)
add 1679ade fixes #17918; update ruby & jekyll, remove incompatible
plugins (#17927)
add 07b8d7a Fix ElemwiseSum for more than 4 inputs (#17995)
add 0597f87 updating stash regex and copy command to fix nightly
imagenet_inference (#18021)
add 58911b5 fixing hyperlinks in python tutorial (#17929)
add c9f8caa Update 3rdparty/mkldnn remote URL and pin to v1.3 (#17972)
add 8d065cc [Numpy] FFI Invocation for Unary Ops (#17779)
add 6692d2c [Bug Fix] support multiple-dim input for unravel_index
(#17748)
add 37c9dd6 Fix for handling negative indices in the fusion of slice
(#17937)
add a1fa6a8 Workaround gnu_tls handshake error on Ubuntu 14.04 Nvidia
Docker (#18018)
add 7dd7e7e Fix issue of zeros gradients w.r.t. RNN bias when num_layers
> 1 (#17872)
add 02ac75e Fix typo in crash course website (#18010)
add 0bff90d Support projection feature of LSTM (#17996)
add 8f6d116 [MKL-DNN] BatchNormRelu Fusion (#17679)
add 7a59239 Remove unused files in Website doc (#16722)
add e3d7866 [Numpy] FFI: random.choice, take and clip (#17854)
add 8e3f0f3 * impl - linalg matrix_rank for cpu and gpu implemented
(#18020)
add a044744 [Numpy] FFI for linalg.qr and linalg.lstsq (#18040)
add f3cfaf9 ffi random (#18051)
add fb73a17 Switch to C++17 and modernize toolchain + CI (#17984)
add e796ae9 Integrate Horovod training API as part of MXNet native
distributed training API (#17531)
add c7d2b3c [NumPy] Add NumPy support for triu (#17614)
add 2c4732b Fix CI (#18056)
add ce48a9d Remove code owner (#17928)
add 94f235d [numpy] add new ffi for column_stack and hstack (#17831)
add 37dbbd4 Fix CD (#18072)
add 5c768f0 [Numpy Extension] Add stop_gradient to npx (#18076)
add afae030 No tensor cores for fp32 interleaved attention, remove div by
8 restriction (#17994)
add 9337137 For mxnet-validation pipeline, require sanity build to
complete successfully before running other build pipelines. (#17999)
add 7bef85e [Numpy] Add ffi for np.sum, np.std, np.var, np.average and
np.histogram (#17866)
add bd0816e Add np.linalg.qr backward (#18050)
add 5155095 add zero grad for npi_unique (#18080)
add b7d1c69 [Numpy] add new ffi for np.linalg.norm (#18066)
add cf93bdc [Numpy] New FFIs for Operator: tile, trace, transpose (#18017)
add b01d1dc Add gelu fuse ops (#18082)
add 4a244dc CI/CD: Remove cuda 9.0, 9.1 (#18087)
add f882de0 [Numpy] add cross product op (#17637)
add 5542d03 Fix and optimize handling of vectorized memory accesses
(#17767)
add dcada9b [Numpy] FFI: random.shuffle, equal, not_equal, less_equal,
greater_equal, less, maximum and minimum (#17896)
add 586c8ab CI: Simplify CentOS7 CI/CD config (#18093)
add 1c0416c Disable MKL_USE_STATIC_LIBS by default
This update added new revisions after undoing existing revisions.
That is to say, some revisions that were in the old version of the
branch are not in the new version. This situation occurs
when a user --force pushes a change and generates a repository
containing something like this:
* -- * -- B -- O -- O -- O (8a9599b)
\
N -- N -- N refs/heads/leezu-patch-1 (1c0416c)
You should already have received notification emails for all of the O
revisions, and so the following emails describe only the N revisions
from the common base, B.
Any revisions marked "omit" are not gone; other references still
refer to them. Any revisions marked "discard" are gone forever.
No new revisions were added by this update.
Summary of changes:
.github/workflows/os_x_staticbuild.yml | 3 +-
.gitmodules | 2 +-
3rdparty/dmlc-core | 2 +-
3rdparty/mkldnn | 2 +-
3rdparty/mshadow/guide/Makefile | 2 +-
3rdparty/mshadow/guide/mshadow-ps/Makefile | 2 +-
3rdparty/mshadow/make/mshadow.mk | 4 +-
3rdparty/mshadow/mshadow/base.h | 72 -
3rdparty/mshadow/mshadow/half2.h | 143 --
3rdparty/mshadow/mshadow/logging.h | 5 +
3rdparty/mshadow/mshadow/packet-inl.h | 4 +
3rdparty/mshadow/mshadow/random.h | 118 +-
3rdparty/mshadow/mshadow/tensor_cpu-inl.h | 9 +-
3rdparty/mshadow/test/Makefile | 2 +-
CMakeLists.txt | 61 +-
CODEOWNERS | 26 +-
Makefile | 34 +-
amalgamation/Makefile | 7 +-
amalgamation/amalgamation.py | 2 +-
benchmark/opperf/README.md | 4 +-
.../nd_operations/array_manipulation_operators.py | 8 +-
benchmark/opperf/rules/default_params.py | 1 -
benchmark/python/ffi/benchmark_ffi.py | 87 ++
cd/Jenkinsfile_cd_pipeline | 2 +-
cd/Jenkinsfile_release_job | 2 +-
cd/README.md | 1 -
cd/mxnet_lib/dynamic/Jenkins_pipeline.groovy | 4 +-
cd/mxnet_lib/mxnet_lib_pipeline.groovy | 10 +-
cd/mxnet_lib/static/Jenkins_pipeline.groovy | 9 +-
cd/python/docker/Jenkins_pipeline.groovy | 5 +-
cd/python/pypi/Jenkins_pipeline.groovy | 5 +-
cd/python/pypi/pypi_package.sh | 2 +-
cd/utils/artifact_repository.md | 4 +-
cd/utils/artifact_repository.py | 6 +-
cd/utils/docker_tag.sh | 4 +-
cd/utils/mxnet_base_image.sh | 6 -
ci/README.md | 95 +-
ci/build.py | 11 +-
ci/dev_menu.py | 2 +-
ci/docker/Dockerfile.build.android_armv7 | 94 +-
ci/docker/Dockerfile.build.android_armv8 | 92 +-
ci/docker/Dockerfile.build.armv6 | 45 +-
ci/docker/Dockerfile.build.armv7 | 54 +-
ci/docker/Dockerfile.build.armv8 | 56 +-
ci/docker/Dockerfile.build.jetson | 96 +-
ci/docker/Dockerfile.build.test.arm_qemu | 47 -
....ubuntu1404_cpu => Dockerfile.build.test.armv7} | 24 +-
....ubuntu1404_cpu => Dockerfile.build.test.armv8} | 24 +-
ci/docker/Dockerfile.build.ubuntu_build_cuda | 11 +-
ci/docker/Dockerfile.build.ubuntu_cpu | 8 +-
ci/docker/Dockerfile.build.ubuntu_cpu_jekyll | 10 +-
ci/docker/Dockerfile.build.ubuntu_cpu_julia | 8 +-
ci/docker/Dockerfile.build.ubuntu_cpu_r | 3 +
ci/docker/Dockerfile.build.ubuntu_cpu_scala | 3 +
ci/docker/Dockerfile.build.ubuntu_gpu_cu100 | 84 --
ci/docker/Dockerfile.build.ubuntu_gpu_cu101 | 14 +-
ci/docker/Dockerfile.build.ubuntu_gpu_cu102 | 85 --
ci/docker/Dockerfile.build.ubuntu_gpu_cu80 | 79 --
ci/docker/Dockerfile.build.ubuntu_gpu_cu90 | 85 --
ci/docker/Dockerfile.build.ubuntu_gpu_cu92 | 84 --
ci/docker/Dockerfile.build.ubuntu_nightly_cpu | 11 +-
ci/docker/Dockerfile.build.ubuntu_nightly_gpu | 11 +-
....centos7_cpu => Dockerfile.publish.centos7_cpu} | 9 +-
...s7_gpu => Dockerfile.publish.centos7_gpu_cu100} | 17 +-
...s7_gpu => Dockerfile.publish.centos7_gpu_cu101} | 17 +-
...s7_gpu => Dockerfile.publish.centos7_gpu_cu102} | 17 +-
...os7_gpu => Dockerfile.publish.centos7_gpu_cu92} | 15 +-
ci/docker/Dockerfile.publish.test.centos7_cpu | 12 +-
ci/docker/Dockerfile.publish.test.centos7_gpu | 12 +-
ci/docker/Dockerfile.publish.test.ubuntu1404_gpu | 40 -
ci/docker/Dockerfile.publish.ubuntu1404_gpu | 39 -
ci/docker/install/android_armv7_openblas.sh | 31 -
ci/docker/install/android_ndk.sh | 38 -
ci/docker/install/arm64_openblas.sh | 35 -
ci/docker/install/centos7_base.sh | 46 -
ci/docker/install/centos7_ccache.sh | 8 +-
ci/docker/install/centos7_core.sh | 10 +
.../install/{arm_openblas.sh => centos7_nccl.sh} | 19 +-
ci/docker/install/deb_ubuntu_ccache.sh | 26 +-
ci/docker/install/{arm_openblas.sh => thrust.sh} | 17 +-
ci/docker/install/ubuntu_arm_qemu.sh | 37 -
ci/docker/install/ubuntu_arm_qemu_bin.sh | 40 -
ci/docker/install/ubuntu_gcc8.sh | 2 +-
ci/docker/install/ubuntu_publish.sh | 88 --
ci/docker/install/ubuntu_r.sh | 5 +-
ci/docker/install/ubuntu_scala.sh | 26 +-
ci/docker/qemu/README.md | 18 -
ci/docker/qemu/runtime_functions.py | 134 --
ci/docker/qemu/vmcontrol.py | 360 -----
ci/docker/runtime_functions.sh | 277 ++--
.../aarch64-linux-gnu-toolchain.cmake} | 24 +-
.../arm-linux-gnueabihf-toolchain.cmake} | 19 +-
ci/jenkins/Jenkins_steps.groovy | 121 +-
ci/jenkins/Jenkinsfile_centos_cpu | 5 +-
ci/jenkins/Jenkinsfile_centos_gpu | 10 +-
ci/jenkins/Jenkinsfile_clang | 4 +-
ci/jenkins/Jenkinsfile_edge | 7 +-
.../{Jenkinsfile_centos_gpu => Jenkinsfile_full} | 46 +-
ci/jenkins/Jenkinsfile_unix_cpu | 3 -
ci/jenkins/Jenkinsfile_unix_gpu | 2 -
...file_centos_cpu => Jenkinsfile_website_nightly} | 25 +-
ci/publish/Jenkinsfile | 2 +-
ci/publish/README.md | 7 +-
ci/qemu/README.md | 92 --
ci/qemu/copy.sh | 23 -
ci/qemu/init.sh | 23 -
ci/qemu/initrd_modif/inittab | 38 -
ci/qemu/install.sh | 32 -
ci/qemu/mxnet_requirements.txt | 7 -
ci/qemu/preseed.cfg | 68 -
ci/qemu/preseed.sh | 29 -
ci/qemu/run.sh | 33 -
ci/qemu/test_requirements.txt | 3 -
ci/safe_docker_run.py | 2 +-
cmake/Modules/FindNCCL.cmake | 10 +-
cmake/upstream/FindCUDAToolkit.cmake | 205 ++-
config/distribution/linux_cu100.cmake | 3 +-
config/distribution/linux_cu101.cmake | 3 +-
config/distribution/linux_cu102.cmake | 3 +-
config/distribution/linux_cu75.cmake | 35 -
config/distribution/linux_cu80.cmake | 35 -
config/distribution/linux_cu90.cmake | 35 -
config/distribution/linux_cu91.cmake | 35 -
config/distribution/linux_cu92.cmake | 3 +-
cpp-package/example/Makefile | 2 +-
cpp-package/example/example.mk | 4 +-
cpp-package/example/feature_extract/Makefile | 4 +-
cpp-package/example/inference/Makefile | 2 +-
cpp-package/example/inference/inference.mk | 4 +-
.../docker-python/Dockerfile.mxnet.python.gpu.cu90 | 29 -
.../Dockerfile.mxnet.python.gpu.cu90.mkl | 29 -
.../Dockerfile.mxnet.python3.gpu.cu90 | 29 -
.../Dockerfile.mxnet.python3.gpu.cu90.mkl | 29 -
docker/docker-python/README.md | 8 -
docker/docker-python/build_python_dockerfile.sh | 21 +-
.../python/tutorials/deploy/run-on-aws/use_ec2.rst | 6 +-
.../getting-started/crash-course/4-train.md | 2 +-
.../getting-started/crash-course/5-predict.md | 2 +-
.../themes/mx-theme/mxtheme/header_top.html | 17 +
.../themes/mx-theme/mxtheme/layout.html | 48 +
docs/static_site/Makefile | 2 +-
docs/static_site/src/Gemfile | 8 +-
docs/static_site/src/Gemfile.lock | 73 +-
docs/static_site/src/_config_beta.yml | 4 +-
docs/static_site/src/_config_prod.yml | 4 +-
docs/static_site/src/_includes/head.html | 39 +
docs/static_site/src/_includes/header.html | 17 +
docs/static_site/src/pages/api/faq/env_var.md | 8 +
.../src/pages/get_started/windows_setup.md | 4 +-
example/distributed_training-horovod/README.md | 7 +-
.../distributed_training/cifar10_kvstore_hvd.py | 237 ++++
example/extensions/lib_custom_op/relu_lib.cu | 90 +-
example/extensions/lib_custom_op/test_relu.py | 43 +-
example/image-classification/predict-cpp/Makefile | 6 +-
example/multi_threaded_inference/Makefile | 2 +-
.../multi_threaded_inference.cc | 5 +-
example/rnn/large_word_lm/setup.py | 2 +-
include/mxnet/base.h | 12 -
include/mxnet/lib_api.h | 57 +-
include/mxnet/random_generator.h | 8 +
include/mxnet/resource.h | 2 +-
include/mxnet/runtime/ffi_helper.h | 18 +
include/mxnet/runtime/object.h | 1 +
julia/deps/build.jl | 2 +-
julia/src/base.jl | 1 +
make/crosscompile.jetson.mk | 216 ---
make/staticbuild/linux_cu100.mk | 8 +
make/staticbuild/linux_cu101.mk | 8 +
make/staticbuild/linux_cu102.mk | 8 +
make/staticbuild/linux_cu75.mk | 167 ---
make/staticbuild/linux_cu80.mk | 170 ---
make/staticbuild/linux_cu90.mk | 172 ---
make/staticbuild/linux_cu91.mk | 172 ---
make/staticbuild/linux_cu92.mk | 8 +
perl-package/AI-MXNet/t/test_init.t | 5 +-
python/mxnet/_ffi/_cython/convert.pxi | 6 +
python/mxnet/_ffi/node_generic.py | 2 +
python/mxnet/_numpy_op_doc.py | 289 +---
.../mxnet/contrib/onnx/mx2onnx/_op_translations.py | 9 +-
python/mxnet/gluon/contrib/nn/basic_layers.py | 3 +-
python/mxnet/gluon/nn/basic_layers.py | 142 +-
python/mxnet/gluon/rnn/rnn_layer.py | 1 +
python/mxnet/gluon/trainer.py | 1 +
python/mxnet/kvstore/__init__.py | 1 +
python/mxnet/kvstore/horovod.py | 161 +++
python/mxnet/kvstore/kvstore.py | 3 +
python/mxnet/ndarray/numpy/_op.py | 1081 ++++++++++++---
python/mxnet/ndarray/numpy/linalg.py | 226 +++-
python/mxnet/ndarray/numpy/random.py | 192 +--
python/mxnet/ndarray/sparse.py | 2 +-
python/mxnet/ndarray_doc.py | 103 +-
python/mxnet/numpy/fallback.py | 4 -
python/mxnet/numpy/fallback_linalg.py | 8 +-
python/mxnet/numpy/linalg.py | 173 ++-
python/mxnet/numpy/multiarray.py | 741 ++++++++++-
python/mxnet/numpy/random.py | 39 +-
python/mxnet/numpy_dispatch_protocol.py | 5 +
python/mxnet/symbol/numpy/_symbol.py | 544 +++++++-
python/mxnet/symbol/numpy/linalg.py | 173 ++-
python/mxnet/symbol/numpy/random.py | 11 +-
python/mxnet/symbol_doc.py | 215 +--
python/setup.py | 4 +-
.../assembly/src/main/assembly/assembly.xml | 2 +-
.../apache/mxnet/util/NativeLibraryLoader.scala | 2 +-
src/api/_api_internal/_api_internal.cc | 10 +
.../np_det.cc} | 25 +-
.../np_eig.cc} | 44 +-
.../np_eigvals.cc} | 41 +-
.../np_inv.cc} | 25 +-
.../numpy/{np_cumsum.cc => linalg/np_lstsq.cc} | 60 +-
.../np_matrix_rank.cc} | 63 +-
.../numpy/{np_cumsum.cc => linalg/np_norm.cc} | 46 +-
.../{np_tensordot_op.cc => linalg/np_pinv.cc} | 60 +-
.../np_potrf.cc} | 30 +-
.../np_qr.cc} | 28 +-
.../np_slogdet.cc} | 26 +-
.../np_solve.cc} | 27 +-
.../np_tensorinv.cc} | 30 +-
.../np_tensorsolve.cc} | 36 +-
src/api/operator/numpy/np_bincount_op.cc | 4 +-
.../operator/numpy/np_broadcast_reduce_op_value.cc | 105 +-
...np_broadcast_reduce_op_value.cc => np_cross.cc} | 32 +-
src/api/operator/numpy/np_cumsum.cc | 4 +-
src/api/operator/numpy/np_delete_op.cc | 101 ++
...ast_op.cc => np_elemwise_broadcast_logic_op.cc} | 57 +-
src/api/operator/numpy/np_elemwise_broadcast_op.cc | 59 +
.../operator/numpy/np_elemwise_unary_op_basic.cc | 93 ++
src/api/operator/numpy/np_histogram_op.cc | 81 ++
src/api/operator/numpy/np_init_op.cc | 207 ++-
src/api/operator/numpy/np_insert_op.cc | 156 +++
.../numpy/{np_bincount_op.cc => np_interp_op.cc} | 61 +-
...roadcast_reduce_op_value.cc => np_matmul_op.cc} | 35 +-
src/api/operator/numpy/np_matrix_op.cc | 256 +++-
src/api/operator/numpy/np_moments_op.cc | 209 +++
src/api/operator/numpy/np_tensordot_op.cc | 4 +-
...broadcast_reduce_op_value.cc => np_trace_op.cc} | 42 +-
..._broadcast_reduce_op_value.cc => np_triu_op.cc} | 32 +-
src/api/operator/numpy/np_window_op.cc | 79 ++
.../random/{np_laplace_op.cc => np_choice_op.cc} | 57 +-
.../{np_cumsum.cc => random/np_exponential_op.cc} | 58 +-
src/api/operator/numpy/random/np_laplace_op.cc | 2 +-
.../operator/numpy/random/np_location_scale_op.cc | 150 +++
.../numpy/{np_cumsum.cc => random/np_pareto_op.cc} | 57 +-
.../numpy/{np_cumsum.cc => random/np_power_op.cc} | 57 +-
.../{np_cumsum.cc => random/np_rayleigh_op.cc} | 57 +-
.../{np_cumsum.cc => random/np_weibull_op.cc} | 57 +-
src/api/operator/random/np_gamma_op.cc | 108 ++
src/api/operator/random/np_normal_op.cc | 96 ++
src/api/operator/random/np_uniform_op.cc | 96 ++
.../shuffle_op.cc} | 39 +-
.../elemwise_binary_broadcast_op_extended.cc} | 35 +-
.../{numpy/np_cumsum.cc => tensor/indexing_op.cc} | 67 +-
src/api/operator/tensor/matrix_op.cc | 92 ++
src/api/operator/ufunc_helper.cc | 24 +-
src/api/operator/ufunc_helper.h | 15 +-
src/api/operator/utils.cc | 5 +
src/api/operator/utils.h | 15 +-
src/c_api/c_api.cc | 41 +-
src/c_api/c_api_executor.cc | 2 +-
src/common/cuda_vectorization.cuh | 283 ++++
src/common/random_generator.cu | 5 +
src/executor/pointwise_fusion_pass.cc | 14 +
src/operator/c_lapack_api.cc | 32 +
src/operator/c_lapack_api.h | 136 +-
src/operator/contrib/batch_norm_relu.cc | 298 +++++
src/operator/contrib/dgl_graph.cc | 73 +-
src/operator/contrib/transformer.cu | 83 +-
src/operator/fusion/fused_op-inl.h | 31 +-
src/operator/fusion/fused_op.cu | 45 +-
src/operator/linalg_impl.h | 53 +-
src/operator/mshadow_op.h | 67 -
src/operator/nn/batch_norm.cc | 8 +-
src/operator/nn/dropout-inl.h | 11 +-
src/operator/nn/dropout.cc | 1 +
src/operator/nn/mkldnn/mkldnn_act.cc | 41 +-
src/operator/nn/mkldnn/mkldnn_base-inl.h | 41 +-
src/operator/nn/mkldnn/mkldnn_base.cc | 47 +-
src/operator/nn/mkldnn/mkldnn_batch_norm-inl.h | 99 +-
src/operator/nn/mkldnn/mkldnn_convolution.cc | 65 +-
src/operator/nn/mkldnn/mkldnn_fully_connected.cc | 4 +-
src/operator/nn/mkldnn/mkldnn_pooling-inl.h | 59 +-
src/operator/nn/mkldnn/mkldnn_pooling.cc | 305 +++--
src/operator/nn/mkldnn/mkldnn_rnn-inl.h | 73 +-
src/operator/nn/mkldnn/mkldnn_rnn.cc | 241 ++--
src/operator/nn/pooling.cc | 6 +-
src/operator/nn/softmax-inl.h | 2 +-
src/operator/numpy/linalg/np_eig-inl.h | 6 +
src/operator/numpy/linalg/np_eigvals-inl.h | 6 +
src/operator/numpy/linalg/np_lstsq-inl.h | 614 +++++++++
src/operator/numpy/linalg/np_lstsq.cc | 97 ++
.../numpy/{np_trace_op.cu => linalg/np_lstsq.cu} | 15 +-
src/operator/numpy/linalg/np_matrix_rank-inl.h | 449 +++++++
src/operator/numpy/linalg/np_matrix_rank.cc | 165 +++
.../{np_trace_op.cu => linalg/np_matrix_rank.cu} | 16 +-
src/operator/numpy/linalg/np_norm-inl.h | 18 +
src/operator/numpy/linalg/np_pinv-inl.h | 14 +
src/operator/numpy/linalg/np_potrf.cc | 3 +-
src/operator/numpy/linalg/np_qr-inl.h | 613 +++++++++
src/operator/numpy/linalg/np_qr.cc | 113 ++
.../numpy/{np_trace_op.cu => linalg/np_qr.cu} | 21 +-
src/operator/numpy/linalg/np_tensorinv-inl.h | 6 +
src/operator/numpy/linalg/np_tensorsolve-inl.h | 6 +
src/operator/numpy/np_boolean_mask_assign.cc | 2 +-
src/operator/numpy/np_broadcast_reduce_op.h | 48 +-
src/operator/numpy/np_broadcast_reduce_op_value.cc | 22 +-
src/operator/numpy/np_broadcast_reduce_op_value.cu | 4 +-
src/operator/numpy/np_cross-inl.h | 1387 ++++++++++++++++++++
src/operator/numpy/np_cross.cc | 121 ++
src/operator/numpy/{np_trace_op.cu => np_cross.cu} | 16 +-
src/operator/numpy/np_delete_op-inl.h | 14 +
src/operator/numpy/np_elemwise_broadcast_op.cc | 4 +
src/operator/numpy/np_elemwise_broadcast_op.h | 68 +-
src/operator/numpy/np_init_op.cc | 2 +-
src/operator/numpy/np_init_op.cu | 6 +-
src/operator/numpy/np_init_op.h | 31 +
src/operator/numpy/np_insert_op-inl.h | 16 +
src/operator/numpy/np_interp_op-inl.h | 286 ++++
src/operator/numpy/np_interp_op.cc | 92 ++
.../numpy/{np_trace_op.cu => np_interp_op.cu} | 18 +-
src/operator/numpy/np_matrix_op-inl.h | 218 +++
src/operator/numpy/np_matrix_op.cc | 132 +-
src/operator/numpy/np_matrix_op.cu | 11 +-
src/operator/numpy/np_trace_op-inl.h | 10 +
src/operator/numpy/np_trace_op.cc | 6 +-
src/operator/numpy/np_trace_op.cu | 4 +-
src/operator/numpy/np_triu_op-inl.h | 241 ++++
src/operator/numpy/np_triu_op.cc | 61 +
.../numpy/{np_trace_op.cu => np_triu_op.cu} | 16 +-
src/operator/numpy/np_unique_op.cc | 1 +
src/operator/numpy/np_where_op-inl.h | 8 +-
src/operator/numpy/np_window_op.h | 8 +
src/operator/numpy/random/np_choice_op.h | 11 +
src/operator/numpy/random/np_exponential_op.h | 8 +
src/operator/numpy/random/np_gamma_op.h | 12 +
src/operator/numpy/random/np_location_scale_op.h | 10 +
src/operator/numpy/random/np_normal_op.h | 12 +
src/operator/numpy/random/np_pareto_op.h | 9 +
src/operator/numpy/random/np_power_op.h | 13 +
src/operator/numpy/random/np_rayleigh_op.h | 9 +
src/operator/numpy/random/np_uniform_op.h | 12 +
src/operator/numpy/random/np_weibull_op.h | 9 +
.../mkldnn/mkldnn_quantized_pooling.cc | 4 +-
src/operator/quantization/quantized_conv.cc | 97 +-
src/operator/quantization/quantized_pooling.cc | 100 +-
src/operator/random/shuffle_op.cc | 2 +-
src/operator/rnn-inl.h | 29 +-
src/operator/rnn.cc | 8 +-
src/operator/rnn_impl.h | 29 +-
src/operator/subgraph/mkldnn/mkldnn_conv.cc | 9 +-
.../subgraph/mkldnn/mkldnn_conv_property.h | 3 +-
.../subgraph/mkldnn/mkldnn_subgraph_base-inl.h | 2 +-
src/operator/tensor/elemwise_binary_op.cuh | 322 +++++
src/operator/tensor/elemwise_binary_op.h | 206 ++-
src/operator/tensor/elemwise_binary_op_basic.cu | 23 +-
src/operator/tensor/elemwise_binary_scalar_op.cuh | 207 +++
src/operator/tensor/elemwise_binary_scalar_op.h | 75 +-
.../tensor/elemwise_binary_scalar_op_basic.cu | 9 +-
.../tensor/elemwise_binary_scalar_op_extended.cu | 15 +-
src/operator/tensor/elemwise_sum.cu | 112 +-
src/operator/tensor/elemwise_sum.h | 14 +-
src/operator/tensor/elemwise_unary_op.cuh | 127 ++
src/operator/tensor/elemwise_unary_op.h | 56 +-
src/operator/tensor/elemwise_unary_op_basic.cc | 1 +
src/operator/tensor/elemwise_unary_op_basic.cu | 1 +
src/operator/tensor/elemwise_unary_op_pow.cu | 1 +
src/operator/tensor/elemwise_unary_op_trig.cu | 1 +
src/operator/tensor/histogram-inl.h | 42 +-
src/operator/tensor/indexing_op.h | 21 +
src/operator/tensor/init_op.h | 29 +
src/operator/tensor/la_op.h | 6 +
src/operator/tensor/matrix_op-inl.h | 13 +
src/operator/tensor/matrix_op.cc | 1 +
src/operator/tensor/ravel.cc | 12 +-
src/operator/tensor/ravel.h | 19 +-
src/resource.cc | 4 +-
tests/cpp/engine/threaded_engine_test.cc | 14 +-
tests/cpp/operator/mkldnn_test.cc | 2 +-
tests/cpp/thread_safety/thread_safety_test.cc | 16 +-
tests/cpp/unittest.mk | 31 +-
tests/jenkins/run_test_pip_installations.sh | 12 +-
tests/nightly/dist_device_sync_kvstore_horovod.py | 80 ++
tests/nightly/test_distributed_training-gpu.sh | 53 +
tests/python/gpu/test_extensions_gpu.py | 18 +-
tests/python/gpu/test_fusion.py | 29 +-
tests/python/gpu/test_gluon_gpu.py | 21 +
tests/python/gpu/test_kvstore_gpu.py | 1 +
tests/python/mkl/test_mkldnn.py | 71 +-
tests/python/quantization/test_quantization.py | 58 +-
tests/python/unittest/test_gluon_data.py | 5 -
tests/python/unittest/test_gluon_rnn.py | 194 ++-
tests/python/unittest/test_init.py | 6 +-
.../python/unittest/test_numpy_interoperability.py | 170 ++-
tests/python/unittest/test_numpy_ndarray.py | 1 +
tests/python/unittest/test_numpy_op.py | 864 +++++++++++-
tests/python/unittest/test_operator.py | 140 ++
tests/python/unittest/test_optimizer.py | 3 +-
tools/dependencies/README.md | 1 -
tools/dependencies/make_shared_dependencies.sh | 3 +-
tools/dependencies/zmq.sh | 6 +
tools/launch.py | 63 +-
tools/pip/doc/CPU_ADDITIONAL.md | 6 +-
tools/pip/doc/CU100_ADDITIONAL.md | 8 +-
tools/pip/doc/CU101_ADDITIONAL.md | 8 +-
tools/pip/doc/CU102_ADDITIONAL.md | 7 +-
tools/pip/doc/CU75_ADDITIONAL.md | 38 -
tools/pip/doc/CU80_ADDITIONAL.md | 38 -
tools/pip/doc/CU90_ADDITIONAL.md | 45 -
tools/pip/doc/CU92_ADDITIONAL.md | 6 +-
tools/pip/doc/NATIVE_ADDITIONAL.md | 7 +-
tools/pip/doc/PYPI_README.md | 2 +-
tools/pip/setup.py | 13 +-
tools/setup_gpu_build_tools.sh | 120 +-
tools/staticbuild/README.md | 2 +-
tools/staticbuild/build.sh | 2 +-
tools/staticbuild/build_lib.sh | 10 +-
tools/staticbuild/build_lib_cmake.sh | 10 +-
416 files changed, 16978 insertions(+), 7495 deletions(-)
delete mode 100755 3rdparty/mshadow/mshadow/half2.h
delete mode 100644 ci/docker/Dockerfile.build.test.arm_qemu
rename ci/docker/{Dockerfile.publish.ubuntu1404_cpu =>
Dockerfile.build.test.armv7} (72%)
rename ci/docker/{Dockerfile.publish.test.ubuntu1404_cpu =>
Dockerfile.build.test.armv8} (72%)
delete mode 100644 ci/docker/Dockerfile.build.ubuntu_gpu_cu100
delete mode 100644 ci/docker/Dockerfile.build.ubuntu_gpu_cu102
delete mode 100644 ci/docker/Dockerfile.build.ubuntu_gpu_cu80
delete mode 100644 ci/docker/Dockerfile.build.ubuntu_gpu_cu90
delete mode 100644 ci/docker/Dockerfile.build.ubuntu_gpu_cu92
copy ci/docker/{Dockerfile.publish.test.centos7_cpu =>
Dockerfile.publish.centos7_cpu} (85%)
copy ci/docker/{Dockerfile.publish.test.centos7_gpu =>
Dockerfile.publish.centos7_gpu_cu100} (74%)
copy ci/docker/{Dockerfile.publish.test.centos7_gpu =>
Dockerfile.publish.centos7_gpu_cu101} (74%)
copy ci/docker/{Dockerfile.publish.test.centos7_gpu =>
Dockerfile.publish.centos7_gpu_cu102} (74%)
copy ci/docker/{Dockerfile.publish.test.centos7_gpu =>
Dockerfile.publish.centos7_gpu_cu92} (77%)
delete mode 100644 ci/docker/Dockerfile.publish.test.ubuntu1404_gpu
delete mode 100644 ci/docker/Dockerfile.publish.ubuntu1404_gpu
delete mode 100755 ci/docker/install/android_armv7_openblas.sh
delete mode 100755 ci/docker/install/android_ndk.sh
delete mode 100755 ci/docker/install/arm64_openblas.sh
delete mode 100755 ci/docker/install/centos7_base.sh
copy ci/docker/install/{arm_openblas.sh => centos7_nccl.sh} (53%)
rename ci/docker/install/{arm_openblas.sh => thrust.sh} (75%)
delete mode 100755 ci/docker/install/ubuntu_arm_qemu.sh
delete mode 100755 ci/docker/install/ubuntu_arm_qemu_bin.sh
delete mode 100755 ci/docker/install/ubuntu_publish.sh
delete mode 100644 ci/docker/qemu/README.md
delete mode 100755 ci/docker/qemu/runtime_functions.py
delete mode 100644 ci/docker/qemu/vmcontrol.py
rename ci/docker/{install/android_arm64_openblas.sh =>
toolchains/aarch64-linux-gnu-toolchain.cmake} (64%)
mode change 100755 => 100644
rename ci/docker/{install/ubuntu_arm.sh =>
toolchains/arm-linux-gnueabihf-toolchain.cmake} (65%)
mode change 100755 => 100644
copy ci/jenkins/{Jenkinsfile_centos_gpu => Jenkinsfile_full} (50%)
copy ci/jenkins/{Jenkinsfile_centos_cpu => Jenkinsfile_website_nightly} (75%)
delete mode 100644 ci/qemu/README.md
delete mode 100755 ci/qemu/copy.sh
delete mode 100755 ci/qemu/init.sh
delete mode 100644 ci/qemu/initrd_modif/inittab
delete mode 100755 ci/qemu/install.sh
delete mode 100644 ci/qemu/mxnet_requirements.txt
delete mode 100644 ci/qemu/preseed.cfg
delete mode 100755 ci/qemu/preseed.sh
delete mode 100755 ci/qemu/run.sh
delete mode 100644 ci/qemu/test_requirements.txt
delete mode 100644 config/distribution/linux_cu75.cmake
delete mode 100644 config/distribution/linux_cu80.cmake
delete mode 100644 config/distribution/linux_cu90.cmake
delete mode 100644 config/distribution/linux_cu91.cmake
delete mode 100644 docker/docker-python/Dockerfile.mxnet.python.gpu.cu90
delete mode 100644 docker/docker-python/Dockerfile.mxnet.python.gpu.cu90.mkl
delete mode 100644 docker/docker-python/Dockerfile.mxnet.python3.gpu.cu90
delete mode 100644 docker/docker-python/Dockerfile.mxnet.python3.gpu.cu90.mkl
create mode 100644 example/distributed_training/cifar10_kvstore_hvd.py
delete mode 100644 make/crosscompile.jetson.mk
delete mode 100644 make/staticbuild/linux_cu75.mk
delete mode 100644 make/staticbuild/linux_cu80.mk
delete mode 100644 make/staticbuild/linux_cu90.mk
delete mode 100644 make/staticbuild/linux_cu91.mk
create mode 100644 python/mxnet/kvstore/horovod.py
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_det.cc} (67%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_eig.cc} (55%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_eigvals.cc} (56%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_inv.cc} (67%)
copy src/api/operator/numpy/{np_cumsum.cc => linalg/np_lstsq.cc} (51%)
copy src/api/operator/numpy/{np_tensordot_op.cc => linalg/np_matrix_rank.cc}
(51%)
copy src/api/operator/numpy/{np_cumsum.cc => linalg/np_norm.cc} (63%)
copy src/api/operator/numpy/{np_tensordot_op.cc => linalg/np_pinv.cc} (51%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_potrf.cc} (67%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_qr.cc} (67%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_slogdet.cc} (67%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_solve.cc} (62%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_tensorinv.cc} (67%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
linalg/np_tensorsolve.cc} (58%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc => np_cross.cc}
(63%)
create mode 100644 src/api/operator/numpy/np_delete_op.cc
copy src/api/operator/numpy/{np_elemwise_broadcast_op.cc =>
np_elemwise_broadcast_logic_op.cc} (52%)
create mode 100644 src/api/operator/numpy/np_elemwise_unary_op_basic.cc
create mode 100644 src/api/operator/numpy/np_histogram_op.cc
create mode 100644 src/api/operator/numpy/np_insert_op.cc
copy src/api/operator/numpy/{np_bincount_op.cc => np_interp_op.cc} (55%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
np_matmul_op.cc} (62%)
create mode 100644 src/api/operator/numpy/np_moments_op.cc
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc =>
np_trace_op.cc} (64%)
copy src/api/operator/numpy/{np_broadcast_reduce_op_value.cc => np_triu_op.cc}
(59%)
create mode 100644 src/api/operator/numpy/np_window_op.cc
copy src/api/operator/numpy/random/{np_laplace_op.cc => np_choice_op.cc} (62%)
copy src/api/operator/numpy/{np_cumsum.cc => random/np_exponential_op.cc} (50%)
create mode 100644 src/api/operator/numpy/random/np_location_scale_op.cc
copy src/api/operator/numpy/{np_cumsum.cc => random/np_pareto_op.cc} (51%)
copy src/api/operator/numpy/{np_cumsum.cc => random/np_power_op.cc} (51%)
copy src/api/operator/numpy/{np_cumsum.cc => random/np_rayleigh_op.cc} (50%)
copy src/api/operator/numpy/{np_cumsum.cc => random/np_weibull_op.cc} (51%)
create mode 100644 src/api/operator/random/np_gamma_op.cc
create mode 100644 src/api/operator/random/np_normal_op.cc
create mode 100644 src/api/operator/random/np_uniform_op.cc
copy src/api/operator/{numpy/np_broadcast_reduce_op_value.cc =>
random/shuffle_op.cc} (63%)
copy src/api/operator/{numpy/np_broadcast_reduce_op_value.cc =>
tensor/elemwise_binary_broadcast_op_extended.cc} (56%)
copy src/api/operator/{numpy/np_cumsum.cc => tensor/indexing_op.cc} (50%)
create mode 100644 src/api/operator/tensor/matrix_op.cc
create mode 100644 src/common/cuda_vectorization.cuh
create mode 100644 src/operator/contrib/batch_norm_relu.cc
create mode 100644 src/operator/numpy/linalg/np_lstsq-inl.h
create mode 100644 src/operator/numpy/linalg/np_lstsq.cc
copy src/operator/numpy/{np_trace_op.cu => linalg/np_lstsq.cu} (74%)
create mode 100644 src/operator/numpy/linalg/np_matrix_rank-inl.h
create mode 100644 src/operator/numpy/linalg/np_matrix_rank.cc
copy src/operator/numpy/{np_trace_op.cu => linalg/np_matrix_rank.cu} (69%)
create mode 100644 src/operator/numpy/linalg/np_qr-inl.h
create mode 100644 src/operator/numpy/linalg/np_qr.cc
copy src/operator/numpy/{np_trace_op.cu => linalg/np_qr.cu} (71%)
create mode 100644 src/operator/numpy/np_cross-inl.h
create mode 100644 src/operator/numpy/np_cross.cc
copy src/operator/numpy/{np_trace_op.cu => np_cross.cu} (72%)
create mode 100644 src/operator/numpy/np_interp_op-inl.h
create mode 100644 src/operator/numpy/np_interp_op.cc
copy src/operator/numpy/{np_trace_op.cu => np_interp_op.cu} (73%)
create mode 100644 src/operator/numpy/np_triu_op-inl.h
create mode 100644 src/operator/numpy/np_triu_op.cc
copy src/operator/numpy/{np_trace_op.cu => np_triu_op.cu} (73%)
create mode 100644 src/operator/tensor/elemwise_binary_op.cuh
create mode 100644 src/operator/tensor/elemwise_binary_scalar_op.cuh
create mode 100644 src/operator/tensor/elemwise_unary_op.cuh
create mode 100644 tests/nightly/dist_device_sync_kvstore_horovod.py
create mode 100755 tests/nightly/test_distributed_training-gpu.sh
delete mode 100644 tools/pip/doc/CU75_ADDITIONAL.md
delete mode 100644 tools/pip/doc/CU80_ADDITIONAL.md
delete mode 100644 tools/pip/doc/CU90_ADDITIONAL.md
