You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@mxnet.apache.org by zh...@apache.org on 2019/05/25 06:44:00 UTC
[incubator-mxnet] branch v1.5.x updated (f9dbd0e -> 136a5df)
This is an automated email from the ASF dual-hosted git repository.
zhasheng pushed a change to branch v1.5.x
in repository https://gitbox.apache.org/repos/asf/incubator-mxnet.git.
omit f9dbd0e Fix crash in random.shuffle operator (#15041)
omit e95b551 Add primitive cache for MKL-DNN sum(elemwise_add operator (#14914)
add c5265fb Add primitive cache for MKL-DNN sum(elemwise_add operator (#14914)
add eecf07a Update MKL-DNN submodule to v0.19 (#14783)
add 60342d6 add KEY for zachgk (#14965)
add d87bd2a Fix the incorrect MKLDNN/MKL logic in cmake (#14877)
add 8d6ac4a Support 3D input for MKL-DNN softmax operator (#14818)
add 0d77947 [MKLDNN] Enable more convolution + activation fusion (#14819)
add 016b3b2 [Dependency Update] Upgrade CI to use latest cuDNN (#14950)
add 5b4e866 Sync Horovod distributed training examples with latest changes (#14748)
add 5aa62d8 Use env var to enforce safe accumulation in ReduceAxesCompute (#14830)
add 379d11f Disables large tensor size cpu test step (#14982)
add c8ee1b3 Fix test memory with ResourceScope (#14666)
add e11a23d [Doc] Add MKL-DNN operator list (#14891)
add 3d6646a Silence excessive mkldnn logging output on tests. (#14947)
add 9f451fb [MXNET-1333] Estimator and Fit API (#14629)
add a7e7cdc add ctx for rand_ndarray and rand_sparse_ndarray (#14966)
add b377130 implementation for equivalence of tf.moments (#14842)
add 8c4a0e6 Disables test_bulking due to flakyness (#14971)
add 0e570a7 Disables flaky test_operator_gpu.test_activation (#14969)
add bd44ff4 Update mkldnn window build instructions in MKLDNN_README.md (#14952)
add 07a8903 clean up duplicate cudnn installation (#14996)
add 720aa29 add an operator for computing the likelihood of a Hawkes self-exciting process (#14683)
add 96b1cde [MXNET-1403] Disable numpy's writability of NDArray once it is zero-copied to MXNet (#14948)
add d84c61c Fixes cuDNN version for CUDA 9.0 build environment (#15001)
add 3cbfe48 Add matrix inversion operator in linalg (#14963)
add 807e40a [MXNET-1402] MXNet docs change for 1.4.1 release (#14949)
add ace478f bump up cudnn to 7.5.1 & nccl 2.4.2 (#14988)
add aac3cdb Revert "[MXNET-1333] Estimator and Fit API (#14629)" (#15008)
add 03e059f [MXNet-1334][Fit API]base class for estimator and eventhandler (#14346)
add e58a001 Fixed issue where the estimator was printing beyond the dataset size … (#14464)
add faec0a9 [MXNet-1349][Fit API]Add validation support and unit tests for fit() API (#14442)
add 767e34d [MXNet-1340][Fit API]Update train stats (#14494)
add c6afb73 [MXNet-1375][Fit API]Added RNN integration test for fit() API (#14547)
add ab7039e [MXNet-1343][Fit API]Add CNN integration test for fit() API (#14405)
add d4f7744 [MXNET-1344, 1346][FIT API] Retrieve Batch size and Logging verbose support for Gluon fit() API (#14587)
add ce21814 move estimator to contrib (#14633)
add 7ae7cb1 move to gluon contrib (#14635)
add 47b65d0 [Fit API] improve event handlers (#14685)
add 40ee94f [MXNET-1396][Fit-API] Update default handler logic (#14765)
add 780ad04 [Fit API] update estimator (#14849)
add 0ac48b6 [Fit-API] Adress PR comments (#14885)
add 96dd059 Fix test randint (#14990)
add 8ae24bf LayerNorm acceleration on GPU (#14935)
add 1e6a1ab update the numpy version
add 2f50c33 add missing numpy version update
add bbbb64a fix the typo
add 35d9c70 missing pylint upgrade
add 95ffa2b add python2 pylint
add 657e699 update the dmlc-core commit to have lint.py use python3
add d9c6ed7 delete the unused python2 pylint
add fbd9df1 add upper bound
add 2e827ab pylint fix
add 7b48c24 Merge pull request #14887 from stu1130/publish_cuda10_1
add 91e8115 Disables flaky test_l2_normalization (#15006)
add 6dd2dda Disables flaky test_droupout (#15003)
add 2cc822a [MKLDNN] Remove repeat parts in MKLDNN.md (#14995)
add 789a6be Add STL checks via -D_GLIBCXX_ASSERTIONS in debug mode (#14896)
add 5bc08ce MXNet AMP (automatic mixed precision) (#14173)
add f680255 bump up atol for test_bilinear_resize_op (#15011)
add aa55e3d Fixes lint issue in AMP (#15015)
add 5854b98 Fix infer shape partial after unknown shape changed to -1 (#14869)
add 90b66b8 [MXNET-1406] [BUG] Fix DLManagedTensor deleter (#15016)
add e5316b1 [MXNET-1401] adding more operators to test support for Large Tensor (#14944)
add 81b4ce2 fix link (#15036)
add cca9c9b use MXNET_SAFE_ACCUMULATION for softmax accumulator (#15037)
add b3c91bf Safe LayerNorm (#15002)
add 37f5315 Disables flaky test_random_size_crop (#15019)
add b0be6c5 Integrating the MKL VML functions to MXNET to speed-up the (element-wised) mathematic computation (#14893)
add d4e458e [Quantization] Support zero-size tensor input for quantization flow (#15031)
add 038b9fb Simplify creation of NodeEntry instances and use emplace_back (#14095)
add d2e397a Add warning for fp16 inputs with MXNET_SAFE_ACCUMULATION=0 (#15046)
add 503c750 Fix warning / static function in header. (#14900)
add 66aa983 Fix crash in random.shuffle operator (#15041)
add eade26f Add cpu implementation for Deformable Convolution (#14879)
add 5763ba9 Decreases test sensitivity (#15014)
add 93fdcad [C++] fix type inconsistent issue when loading quantized parameters (#15038)
add eb0b8af Add cpu implementation for Deformable PSROIPooling (#14886)
add d0ff3cd [MXNET-545] Fix broken cython build (#10951)
add 653cbb4 MKLDNN RNN Inference Integration(fp32 LSTM and vRNN with tanh and relu) (#14713)
add db2295b Allow clearing gpu cache (#14252)
add 9250a73 Revert the change broadcast_to param shape (#14998)
add 136a5df GPU RNN to use TempSpace resource for workspace. (#15056)
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 (f9dbd0e)
\
N -- N -- N refs/heads/v1.5.x (136a5df)
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:
3rdparty/dmlc-core | 2 +-
3rdparty/mkldnn | 2 +-
3rdparty/mshadow | 2 +-
3rdparty/tvm | 2 +-
CMakeLists.txt | 14 +-
CONTRIBUTORS.md | 4 +-
KEYS | 59 ++
Makefile | 4 +-
NEWS.md | 18 +
README.md | 1 +
ci/Jenkinsfile_utils.groovy | 4 +-
ci/build_windows.py | 30 +
ci/docker/Dockerfile.build.centos7_gpu | 2 +-
ci/docker/Dockerfile.build.ubuntu_base_gpu | 2 +-
ci/docker/Dockerfile.build.ubuntu_build_cuda | 4 +-
ci/docker/Dockerfile.build.ubuntu_gpu_cu100 | 3 +-
ci/docker/Dockerfile.build.ubuntu_gpu_cu90 | 2 +-
ci/docker/Dockerfile.build.ubuntu_gpu_cu92 | 2 +-
ci/docker/Dockerfile.build.ubuntu_nightly_gpu | 3 +-
ci/docker/install/docs_requirements | 4 +-
ci/docker/install/ubuntu_mklml.sh | 2 +-
ci/docker/install/ubuntu_publish.sh | 4 +-
ci/docker/install/ubuntu_python.sh | 4 +-
ci/docker/runtime_functions.sh | 80 ++-
ci/jenkins/Jenkins_steps.groovy | 70 +-
ci/jenkins/Jenkinsfile_windows_cpu | 5 +-
ci/qemu/mxnet_requirements.txt | 2 +-
ci/windows/test_py2_cpu.ps1 | 4 +
ci/windows/test_py2_gpu.ps1 | 4 +
ci/windows/test_py3_cpu.ps1 | 4 +
ci/windows/test_py3_gpu.ps1 | 4 +
cmake/BuildCythonModules.cmake | 38 ++
cmake/ChooseBlas.cmake | 16 +-
cmake/DownloadMKLML.cmake | 10 +-
cmake/Modules/FindMKL.cmake | 51 +-
.../clojure-package/test/good-test-ndarray-api.clj | 2 +-
.../clojure-package/test/good-test-symbol-api.clj | 2 +-
cpp-package/example/test_ndarray_copy.cpp | 62 ++
cpp-package/include/mxnet-cpp/ndarray.h | 9 +-
cpp-package/include/mxnet-cpp/ndarray.hpp | 15 +-
cpp-package/tests/ci_test.sh | 3 +
docs/_static/js/options.js | 2 +-
docs/_static/mxnet-theme/index.html | 6 +-
docs/api/python/symbol/linalg.md | 1 +
docs/faq/env_var.md | 29 +
docs/faq/new_op.md | 22 +
docs/faq/perf.md | 9 +-
docs/install/download.md | 1 +
docs/install/index.md | 68 +-
docs/install/requirements.txt | 2 +-
docs/tutorials/amp/amp_tutorial.md | 266 ++++++++
docs/tutorials/{vision => amp}/index.md | 0
docs/tutorials/index.md | 2 +
docs/tutorials/mkldnn/MKLDNN_README.md | 595 ++++++++---------
docs/tutorials/mkldnn/operator_list.md | 88 +++
example/distributed_training-horovod/README.md | 22 +-
.../distributed_training-horovod/gluon_mnist.py | 27 +-
.../distributed_training-horovod/module_mnist.py | 26 +-
.../resnet50_imagenet.py | 42 +-
example/quantization/imagenet_gen_qsym_mkldnn.py | 4 +-
example/ssd/quantization.py | 4 +-
include/mxnet/c_api.h | 23 +
include/mxnet/ndarray.h | 55 +-
include/mxnet/storage.h | 8 +
make/config.mk | 3 +
make/pip/{pip_linux_cu91.mk => pip_linux_cu101.mk} | 2 +-
...pip_linux_cu100mkl.mk => pip_linux_cu101mkl.mk} | 3 +-
python/mxnet/context.py | 18 +
.../mxnet/contrib/amp}/__init__.py | 5 +-
python/mxnet/contrib/amp/amp.py | 344 ++++++++++
.../mxnet/contrib/amp/lists}/__init__.py | 6 +-
python/mxnet/contrib/amp/lists/symbol.py | 609 +++++++++++++++++
python/mxnet/contrib/amp/loss_scaler.py | 77 +++
python/mxnet/cython/base.pyi | 31 +-
python/mxnet/cython/ndarray.pyx | 52 +-
python/mxnet/gluon/block.py | 4 +-
.../mxnet/gluon/contrib/estimator}/__init__.py | 8 +-
python/mxnet/gluon/contrib/estimator/estimator.py | 408 ++++++++++++
.../mxnet/gluon/contrib/estimator/event_handler.py | 705 ++++++++++++++++++++
python/mxnet/gluon/contrib/rnn/conv_rnn_cell.py | 12 +-
python/mxnet/gluon/contrib/rnn/rnn_cell.py | 8 +-
python/mxnet/gluon/parameter.py | 3 +-
python/mxnet/gluon/rnn/rnn_cell.py | 20 +-
python/mxnet/gluon/trainer.py | 7 +
python/mxnet/model.py | 12 +-
python/mxnet/module/module.py | 4 +-
python/mxnet/ndarray/ndarray.py | 14 +-
python/mxnet/symbol/symbol.py | 17 +-
python/mxnet/test_utils.py | 70 +-
python/setup.py | 20 +-
.../customop/CustomOpExampleSuite.scala | 57 +-
.../IMClassificationExampleSuite.scala | 42 +-
.../ImageClassifierExampleSuite.scala | 35 +-
.../infer/predictor/PredictorExampleSuite.scala | 38 +-
.../mxnetexamples/multitask/MultiTaskSuite.scala | 18 +-
.../neuralstyle/NeuralStyleSuite.scala | 24 +-
.../apache/mxnetexamples/rnn/ExampleRNNSuite.scala | 56 +-
.../scala/sample/ImageClassificationExample.scala | 2 +-
src/c_api/c_api.cc | 11 +-
src/c_api/c_api_function.cc | 2 +-
src/c_api/c_api_symbolic.cc | 24 +
src/executor/graph_executor.cc | 11 +-
src/executor/infer_graph_attr_pass.cc | 4 +-
src/imperative/cached_op.cc | 69 +-
src/imperative/imperative.cc | 4 +-
src/ndarray/ndarray.cc | 23 +-
src/nnvm/gradient.cc | 28 +-
src/nnvm/legacy_op_util.cc | 25 +-
src/nnvm/plan_memory.cc | 8 +-
src/operator/c_lapack_api.cc | 26 +-
src/operator/c_lapack_api.h | 100 ++-
src/operator/contrib/all_finite-inl.h | 100 +++
src/operator/contrib/all_finite.cc | 168 +++++
src/operator/contrib/all_finite.cu | 107 +++
.../{pooling_v1.cu => contrib/amp_graph_pass.cc} | 58 +-
src/operator/contrib/deformable_convolution-inl.h | 91 ++-
src/operator/contrib/deformable_convolution.cc | 8 +-
.../contrib/deformable_psroi_pooling-inl.h | 50 +-
src/operator/contrib/deformable_psroi_pooling.cc | 328 ++++++++-
src/operator/contrib/deformable_psroi_pooling.cu | 406 ++++++-----
src/operator/contrib/hawkes_ll-inl.h | 506 ++++++++++++++
src/operator/contrib/hawkes_ll.cc | 148 +++++
.../contrib/{sync_batch_norm.cu => hawkes_ll.cu} | 21 +-
src/operator/contrib/nn/deformable_im2col.cuh | 372 ++++++-----
src/operator/contrib/nn/deformable_im2col.h | 398 ++++++++++-
src/operator/custom/custom.cc | 6 +-
src/operator/elemwise_op_common.h | 4 +-
src/operator/instance_norm-inl.h | 2 +-
src/operator/l2_normalization-inl.h | 12 +-
src/operator/l2_normalization.cc | 4 +-
src/operator/linalg.h | 49 ++
src/operator/linalg_impl.h | 254 ++++++-
src/operator/mkl_functions-inl.h | 165 +++++
src/operator/nn/activation.cc | 2 +-
src/operator/nn/batch_norm.cc | 38 +-
src/operator/nn/dropout-inl.h | 18 -
src/operator/nn/dropout.cc | 23 +-
src/operator/nn/layer_norm-inl.h | 141 ++--
src/operator/nn/layer_norm.cc | 76 ++-
src/operator/nn/layer_norm.cu | 717 ++++++++++++++++++++
src/operator/nn/lrn.cc | 2 +-
src/operator/nn/mkldnn/mkldnn_act.cc | 6 +
src/operator/nn/mkldnn/mkldnn_base-inl.h | 5 +-
src/operator/nn/mkldnn/mkldnn_base.cc | 4 +-
src/operator/nn/mkldnn/mkldnn_convolution-inl.h | 28 +-
src/operator/nn/mkldnn/mkldnn_convolution.cc | 52 +-
src/operator/nn/mkldnn/mkldnn_deconvolution.cc | 18 +-
src/operator/nn/mkldnn/mkldnn_rnn_impl.h | 740 +++++++++++++++++++++
src/operator/nn/mkldnn/mkldnn_softmax.cc | 61 +-
src/operator/nn/moments-inl.h | 254 +++++++
src/operator/nn/moments.cc | 85 +++
src/operator/nn/{lrn.cu => moments.cu} | 20 +-
src/operator/nn/pooling.cc | 4 +-
src/operator/nn/softmax-inl.h | 72 +-
src/operator/nn/softmax.cc | 2 +-
src/operator/operator_common.h | 19 +-
src/operator/pooling_v1-inl.h | 4 +-
src/operator/quantization/dequantize-inl.h | 7 +
src/operator/quantization/quantize-inl.h | 11 +-
src/operator/quantization/quantize_graph_pass.cc | 34 +-
src/operator/quantization/quantize_v2-inl.h | 11 +-
src/operator/quantization/quantized_activation.cc | 4 +
.../quantization/quantized_elemwise_add.cc | 4 +
.../quantization/quantized_fully_connected.cc | 18 +-
src/operator/random/shuffle_op.cc | 1 +
src/operator/regression_output-inl.h | 2 +-
src/operator/rnn-inl.h | 220 ++++--
src/operator/rnn.cc | 472 ++++++++++++-
src/operator/rnn_impl.h | 7 +
src/operator/subgraph/build_subgraph.cc | 17 +-
src/operator/subgraph/mkldnn/mkldnn_conv-inl.h | 24 +-
src/operator/subgraph/mkldnn/mkldnn_conv.cc | 71 +-
.../subgraph/mkldnn/mkldnn_conv_property.h | 136 ++--
.../mkldnn/mkldnn_fc_post_quantize_property.h | 7 +
src/operator/subgraph/mkldnn/mkldnn_fc_property.h | 9 +-
.../mkldnn_post_quantize_align_scale_property.h | 10 +-
.../mkldnn/mkldnn_post_quantize_property.h | 7 +
.../subgraph/mkldnn/mkldnn_subgraph_property.cc | 16 +-
src/operator/subgraph/subgraph_property.h | 52 +-
src/operator/tensor/amp_cast.cc | 150 +++++
.../tensor/{control_flow_op.cu => amp_cast.cu} | 23 +-
src/operator/tensor/amp_cast.h | 165 +++++
src/operator/tensor/broadcast_reduce_op.h | 32 +-
src/operator/tensor/broadcast_reduce_op_index.cc | 5 +-
src/operator/tensor/broadcast_reduce_op_value.cc | 12 +-
src/operator/tensor/control_flow_op.cc | 7 +-
src/operator/tensor/control_flow_op.h | 1 +
src/operator/tensor/dot-inl.h | 15 +-
src/operator/tensor/elemwise_binary_broadcast_op.h | 4 +-
src/operator/tensor/elemwise_sum.cc | 11 +-
src/operator/tensor/elemwise_unary_op.h | 123 ++--
src/operator/tensor/elemwise_unary_op_basic.cc | 77 ++-
src/operator/tensor/indexing_op.cc | 12 +-
src/operator/tensor/indexing_op.h | 4 +-
src/operator/tensor/la_op-inl.h | 40 +-
src/operator/tensor/la_op.cc | 50 ++
src/operator/tensor/la_op.cu | 6 +
src/operator/tensor/la_op.h | 15 +
src/operator/tensor/matrix_op-inl.h | 2 +-
src/operator/tensor/ordering_op.cc | 4 +-
src/storage/pooled_storage_manager.h | 6 +-
src/storage/storage.cc | 11 +
src/storage/storage_manager.h | 8 +
tests/cpp/include/test_core_op.h | 4 +-
tests/cpp/operator/mkldnn_test.cc | 2 +-
tests/nightly/JenkinsfileForBinaries | 13 +-
tests/nightly/estimator/test_estimator_cnn.py | 151 +++++
tests/nightly/estimator/test_sentiment_rnn.py | 276 ++++++++
tests/nightly/test_large_array.py | 92 ++-
tests/python/gpu/test_gluon_gpu.py | 1 +
tests/python/gpu/test_operator_gpu.py | 125 +++-
tests/python/mkl/test_subgraph.py | 341 ++++++----
tests/python/quantization/test_quantization.py | 256 +++----
tests/python/tensorrt/test_tensorrt_lenet5.py | 8 +-
tests/python/unittest/test_amp.py | 83 +++
tests/python/unittest/test_contrib_hawkesll.py | 161 +++++
tests/python/unittest/test_gluon_estimator.py | 371 +++++++++++
tests/python/unittest/test_gluon_event_handler.py | 198 ++++++
tests/python/unittest/test_image.py | 1 +
tests/python/unittest/test_infer_shape.py | 95 ++-
tests/python/unittest/test_ndarray.py | 4 +-
tests/python/unittest/test_operator.py | 383 +++++++----
tests/tutorials/test_sanity_tutorials.py | 4 +-
tests/tutorials/test_tutorials.py | 3 +
tools/pip/setup.py | 2 +-
tools/setup_gpu_build_tools.sh | 46 +-
226 files changed, 12744 insertions(+), 2373 deletions(-)
create mode 100644 cmake/BuildCythonModules.cmake
create mode 100644 cpp-package/example/test_ndarray_copy.cpp
create mode 100644 docs/tutorials/amp/amp_tutorial.md
copy docs/tutorials/{vision => amp}/index.md (100%)
create mode 100644 docs/tutorials/mkldnn/operator_list.md
copy make/pip/{pip_linux_cu91.mk => pip_linux_cu101.mk} (99%)
copy make/pip/{pip_linux_cu100mkl.mk => pip_linux_cu101mkl.mk} (99%)
copy {plugin/opencv => python/mxnet/contrib/amp}/__init__.py (93%)
create mode 100755 python/mxnet/contrib/amp/amp.py
copy {plugin/opencv => python/mxnet/contrib/amp/lists}/__init__.py (88%)
create mode 100644 python/mxnet/contrib/amp/lists/symbol.py
create mode 100755 python/mxnet/contrib/amp/loss_scaler.py
copy {plugin/opencv => python/mxnet/gluon/contrib/estimator}/__init__.py (90%)
create mode 100644 python/mxnet/gluon/contrib/estimator/estimator.py
create mode 100644 python/mxnet/gluon/contrib/estimator/event_handler.py
create mode 100755 src/operator/contrib/all_finite-inl.h
create mode 100755 src/operator/contrib/all_finite.cc
create mode 100755 src/operator/contrib/all_finite.cu
copy src/operator/{pooling_v1.cu => contrib/amp_graph_pass.cc} (52%)
create mode 100644 src/operator/contrib/hawkes_ll-inl.h
create mode 100644 src/operator/contrib/hawkes_ll.cc
copy src/operator/contrib/{sync_batch_norm.cu => hawkes_ll.cu} (71%)
mode change 100644 => 100755
create mode 100644 src/operator/mkl_functions-inl.h
create mode 100644 src/operator/nn/mkldnn/mkldnn_rnn_impl.h
create mode 100644 src/operator/nn/moments-inl.h
create mode 100644 src/operator/nn/moments.cc
copy src/operator/nn/{lrn.cu => moments.cu} (74%)
create mode 100644 src/operator/tensor/amp_cast.cc
copy src/operator/tensor/{control_flow_op.cu => amp_cast.cu} (65%)
create mode 100644 src/operator/tensor/amp_cast.h
create mode 100644 tests/nightly/estimator/test_estimator_cnn.py
create mode 100644 tests/nightly/estimator/test_sentiment_rnn.py
create mode 100644 tests/python/unittest/test_amp.py
create mode 100644 tests/python/unittest/test_contrib_hawkesll.py
create mode 100644 tests/python/unittest/test_gluon_estimator.py
create mode 100644 tests/python/unittest/test_gluon_event_handler.py