* Fold affine transformations on 4d convolution

* Handle more cases for affine parameters

* Style fix

* Temporary broadcast optimizations till halide support is merged

* Use computed broadcast axes

* Move shape squeeze

* Review feedback: Add comments

* Patch broadcast axes after the last refactor

* Squeeze input shape

* onnx: add 'constant' operator
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: getting attribute value by name
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: fix code style
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: fix clang compilation warnings
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: exception
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: add 'split' operator
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: add public interface
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: add initial unit test for importer
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: initial implementetion of operator' set
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* [WIP] Unit test for split operation.

* Fix Split Op bounds calculation + UT

* clang format

* Split Op with variable parts unit test.

* Remove unused headers

* General purpose exceptions.

* Change not_supported_error message template.

* Add new general purpose errors.

* ONNX BatchNormalization operation.

* Clang-format

* Update CMake

* Add fixed test data.

* Add missing ngraph install prefix for cmake in travis Dockerfile.

* Remove -Wno-zero-as-null-pointer-constant

* Code review

* Apply clang-format-3.9

* Add missing onnx_import interface files to CMakeList

* Clean code.

* Fix test.

* Apply clang-format-3.9

* Copyright notice format

* Remove inputs in separate files

* use all_close to compare floating point values

* missed changing one CPU to INTERPRETER for unit test

* Add OneHot op to direct execution.

* Remove exceptions from kernel.

* Use Eigen::Tensor.

* Removed unused code.

* format-fix

* Added MKLDNN concat in DEX

* Fix allocation size in batchnorm kernel

* added missing brackets

* Support in-place input & output propagation

* Clarify ng backends as devices

* revise some intros and add menu link

* Initial stab at a trigger script for Jenkins mult-branch pipeline

* CHANGE_URL is not working, so disable for now

* CHANGE_AUTHOR not found either, disabling

* Disable some prints of vars that do not appear to be available

* Add in node block

* Correct the node-label expression

* Put Jenkins repo in a subdirectory and use the requested branch

* Fix groovy syntax error

* Call ngraph-unittest.groovy

* Corrected the branch name for cje-algo

* Now that the GitHub plugin is working, try restoring CHANGE_URL and CHANGE_AUTHOR

* Set an environment variable MB_PIPELINE_CHECKOUT, to trigger proper checkout command in runNgraphBuild()

* Try alternate load mechanism

* I've settled on using "load()" to directly run the code in ngraph-unittest.groovy, so remove vestigal loading of an ngraphUnitTest() function

* Change parameter assignments to be of the form params.var = 'new value'

* Try alternate method of setting parameter values for ngraph-unittest.groovy

* Try parameters-setting syntax from https://qa.nuxeo.org/jenkins/pipeline-syntax/globals#params

* Correct typo

* Try to work around parameter expansion issue

* Modify strategy for passing parameters and using MB Pipeline style checkouts, as per discussions with Nick

* Fixed calling syntax

* Simply call

* Make job parameters as local variables and passed as parameters to ngraphCIPremerge() function, as many of these are used in the closure for parallel tasks, and the lexically scoped global vaiariables was causing function signature mismatches on runNgraphBuild() calls.

* Remove prints

* Move CI to working from Jenkins cje-algo master branch

* Add standard header copyright message

* Changes from suggestions in PR review

* Revert env.WORKSPACE (requested in a review) because it wasn't working -- a "null" ending up in the load path, for unknown reasons

* enable parameter validation for all unit tests

* Remove validation checks from performance critical code paths and skip layout propagation to inputs

* Add templated call method to backend for cases where users need input validation

* Added missing return

* fix python api compile error due to ngraph api change.

* disable parameter validation in python api

* make validating call a separate call rather than templated

* Update Dockerfile

reverted the INTERPRETER CHANGES

* Added a conditional for a prebuilt argument based on task type

* Added default arg and also conditional cmake argument

* Update .travis.yml

* Update Dockerfile

* remove test that relies on CPU if CPU is not built

* fix docker build

* change onnx to use INTERPRETER

* run unit-test-check

* don't propagate outputs inplace to constants

* apply formatting

* refactor for style-check

* fix typo

* onnx: add 'constant' operator
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: getting attribute value by name
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: fix code style
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: fix clang compilation warnings
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: exception
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: add 'split' operator
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: add public interface
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: add initial unit test for importer
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* onnx: initial implementetion of operator' set
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

* [WIP] Unit test for split operation.

* Fix Split Op bounds calculation + UT

* clang format

* Split Op with variable parts unit test.

* Remove unused headers

* Add missing ngraph install prefix for cmake in travis Dockerfile.

* Remove -Wno-zero-as-null-pointer-constant

* Code review

* Apply clang-format-3.9

* Add missing onnx_import interface files to CMakeList

* Copyright notice format

* Dex non-mkldnn version of clipped relu

* Change to static_cast

* separate backend updates for shared and static libs into separate PR

* add search for shared libs to statically linked libs

* update per review comments

* update per review comments

* Update Dockerfile

Added -j 4 to speed up build times

* Update Dockerfile

j -4 comes across the same error as j -8 so moving this to j -2

* Update Dockerfile

* Update Dockerfile

* experimenting with less number of tasks

* Update Dockerfile

* Added cmake changes to use the INTERPRETER instead of cpu.

* Update .travis.yml

* Add cudnn host data param case for CUDNN_DATA_INT8x32

* Default case.

* break

* def guards

* removed def guard for ci test.

* put def guards back

* Remove default to make clang happy.

* Remove clang default switch warning on cudnnDataType_t enum.

* Formatting.

* Added Wswitch-enum supression.

* Back to switch, remove pragma clang suppression.

* disable one more switch compiler check

* Remove default break after throw.

* Fix allocation size in batchnorm kernel

* added missing brackets

* Support GPUKernelArgs in Elementwise-collective and Nd-Convolution.

* Update op::ReplaceSlice to use GPUKernelArgs and unroll coordinate transform loop.

* Formatting.

* Moved function signature for global kernels back to emitter body.

* Formatting.