* extra *block_size

* change grid_size to threads

* hacking to support dot of 3 by 2 inputs with gemm_batch.

* clean up.

* nbench cleanup

* update style

* tf group convolution

* change perms

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

* onnx: add '\n' at end of files
Signed-off-by: Artur Wojcik <artur.wojcik@intel.com>

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

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

* declare sigmoid for core fusion

* add simple test for sigmoid

* info fusion status

* cp op as main op

* builds as expected

* move sigmoid fusion code

* add reference kernel

* sigmoid bprop reference kernel and clang-format

* add delta to bprop

* fprop called

* compiles bprop

* move tests

* serializer support

* address comments in code

* add doc

* naming similar to core ops

* fix failing test

* fix failing test

* address clang issue

* more changes

* change test macro

* 1. Added MKLDNNN BoundedRelu op support for Relu6
2. CpuLayout && CPU assignment pass for BoundedRelu Op
3. Unit test inter v/s CPU for BoundedReluOp
4. MKLDNN and default emitter code for BoundedReluOp

* Removed Debug prints

* 1. Added support for boundedrelu to work on any constant literal
2. unit test case for rank2, rank3, rank4 for bounded relu without serialized graph

* Removed is_six() method

* Reshape bias to 1D for conv + bias bprop fusion

* Reshape goe2 back to 2D before replacing

* added shape checks to validate conv+bias op.

* removed conv+bias backprop merge for separate PR review.

* fixed conv_bias_bprop test.

* minor changes to error messages.

* add gpu_timer to external function

* compiled version

* working version

* using block_begin and block_end

* add the missing '
;'

* move slice to cuda emiter

* change size_t to uint32_t in kernel

* working version

* change block size from 1 to 64

* fix bugs

* nthreads need to be size_t in broadcast op

* add rank to kernel name hash

* update slice in convolution

* resolve index conflict

* change align to align_to_blocksize, add overflow check

* add gird size check and fix pool merge bug

* code style, change names

* - Fixed replace output for the multi layer recurrent cell state tensor output
- Modified rnn add_output to consider direction and n_layer while calculating the output size for mkldnn dst_layer and dst_iter

* fix unit test failure

* collector

* keeping track of inputs; simplifying a merging stratey; adding LKGraph

* LoopKernel Collector

* address feedback

* address feedback 2

* address feedback 3

* add lambda handler support for logger
* reuse logger function

* Added blank convolution kernel and refactored coordinate transform kernel helper.

* Added op::Reshape to the CUDAEmitter.

* Added 2-Nd tiled convolution.

* Bug fixes with data_dilation and filter loop. Still need to add test for coverage of register tiling.

* Styling.

* Removed some comments and code added for testing.

* Some tests became enabled in merge, removing them.

* IntelGPUBackend: create_tensor

* 9 tests are passes. List updated

* workaround for depthwise convolution

* fixe error msg

* Reshape bias to 1D for conv + bias bprop fusion

* Reshape goe2 back to 2D before replacing

* Reshape 4d

* Support dimshuffles/transpose with MKLDNN

* Addressing PR Feedback

* Use Eigen for 3D dimshuffles

- fixes segfault issue for GNMT model execution through ngraph-mxnet

* enable multi datatpye support for Cudnn. refactor binary ops using cudnn

* fix bugs

* add tests to skip list that CUDNN does not support

* not int support on cudnn for backward pooling

* no GPU.dot_4d_5d_multi_axis_big_fp64_VERY_SLOW test anymore

* clang format

* throw if datatype is int8 or int32 for backward pooling

* comments

* fix list in unit_test.manifest

* add type support for alpha, beta

* fix bugs

* datatype support for alpha, beta

* missing ()

* clang format

* batchnorm backward bug fix

* remove debug info

* change member function name to snake case. remove comments

* use nullptr instead of NULL

* code style, use cuDNN everywhere in comments

* add cudnn host parameters memory manager.

* change name to allocate_by_datatype

* compiled

* debug

* fix bug: using list instead of vector, vector address will change each time it resize

* add CUDNN_DATA_UINT8 and CUDNN_DATA_UINT8x4

* constant broadcast folding

* code review feedback

* Move maxpool and avgpool into CudaKernelBuilder and add cache parameters to kernel name for broadcast which are required for correct lookup.

* Styling.

* Add space before avg_pool.

* add gpu_timer to external function

* compiled version

* working version

* using block_begin and block_end

* add the missing '
;'

* get_get_output_elements

* fix comp error

* address scott's feedback

* group conv fix

* group conv fix

* fix typo

* 1. Added mkldnn support for Softmax
2. layout assignment for mkldnn softmax

* added assert to check softmax axis for mkldnn

* onnx: add importer cmakes
* onnx: use file(DOWNLOAD ...) command to download onnx.proto
* onnx: add Protobuf minimal required version

* cmake runs for interpreter

* more updates towards building on windows

* inplace compute

* fix warnings

* Initial support for convolution sum fusion

* Added in-place support for conv sum fusion and test cases

* reverting spurious changes

* Bug fix to account for inplace input in conv sum fusion

* fix compilation error

* Addressed PR feedback