* reduce fprop cache outputs

* refactor traverse nodes

* Slight refactor, add test, adress PR comments

* fix formatting

* Switch to using mkldnn memory descriptors for layout

* More changes for using mkldnn descriptor instead of format

* Removed mkldnn format from cpu layout descriptor. TODO - shuffle folding

* Rotate mkldnn layouts on transpose

* Modifications to builder reshape to skip rotated layouts

* More fixes to layouts and removes axis order from cpu layout descriptor

* Code cleanup

* Removed shuffle folding pass since the functionality is subsumed by the layout pass

* Canonicalize a few more formats to keep MKLDNN happy.

* Style fixes

* Style fixes

* Style fixes

* Addressed PR feedback and added reshape passthrough for non-transpose cases

* Adjust named formats for weights tensors to keep MKLDNN happy

* Style fixes

* resolved merge issues

* Utilize GPUMemoryManager/Allocator for preallocation of intermediate tensor buffer memory.

* Formatting.

* Merge with master required rework of memory due to CFE pass. Moved function memory pool allocation to pass as a result.

* Formatting.

* Added pass source files.

* Updated tests to account for new assert check. All GPUAllocators should be deconstructed before allocation is made in GPUMemoryManager.

* GPUAllocator::close() can be used to close the allocator prior to destruction

* Removed open allocators. Replaced check with inspection of pass::MemoryManager node list.

* Formatting.

* Rename m_memory_buffers -> m_tensor_memory_buffers. Use full path to static alignment variable.

* FunctionMemoryReservation -> TensorMemoryReservation. Only return true in pass if reservation is made (bug fix).

* Moved static compilation mutex.

* Update external function with new pass name.

* GPU_ExternalFunction: Add s_memory_pool_alignment, remove optimize_and_assemble method.

* lrn init

* fix comment

* mkldnn lrn (#1295)

* add serializer + fix compiler warnings

* updated

* type prop

* disable test in manifest

* try to exclude

* style

* double

* dobule

* more

* style

* more

* vecs

* fix goe

* Testing out some ideas for better error messages on AvgPool

* Add uncaught_exception() check to ConstructionAssertLogger dtor

* More general assertion class, not homed inside Node

* Minor formatting change

* NODE_ASSERT for type prop failure

* Produce lighter-weight DummyAssertionHandler when assertion succeeds

* New ctor for AssertionHelper that takes a single location arg; more const&-ness for the constructors

* Remove move constructor for AssertionHelper; fix broken test in assertion.cpp

* Miscellaneous improvements

* Templatized AssertionHelper so different exception classes can be used; implemented TYPE_CHECK_ASSERT around this
* Changed from a "stack" of locations to a single location (the stack was too complicated)
* Added "FAIL" classes/macros which do not take a condition

* Rename a helper function

* Cleanup, cruft removal

* Add test to make sure the assert helper has the lifetime we expect

* Missing includes

* IntelGPUBackend: Broadcast operation

* IntelGPUBackend: more tests for Broadcast operation

* Move macro to static C function in Broadcast tests

* make pool test check backends other than CPU

* more unit test cleanup

* cpu loop kernel fusion pass

*  remove extra code

* bounded relu test

* address scotts feedback

* CPU Direct Execution: Implement ConvertLayout and refactor

* CPU Direct Execution: Implement Convolution

* 1) Adds computation reuse to direct execution
2) Add avg_pool, broadcast and convolution_bias to direct execution
3) Moved some computation reuse utility functions to graph_utils

* Use lists instead of vectors to avoid reallocation overheads

* - Added convolution variants to direct execution
- Removed ConvolutionBiasRelu, use ConvolutionBias instead
- Reduced code duplication by moving functionality to mkldnn_emitter
  from cpu_emitter

* Style fix

* Moved mkldnn build_convolution to a templated method

* Style fix

* refactored mkldnn conv bprop builders

* Style fix

move long building tests to the be the first tests built with the hope of reducing build time. (#1229)

* get_subgraph_outputs

* simplify the condition

* reshape inplace without copy data if possible.

* added reshape and broadcast to CSE.

* Fixed debug messages.

* open only the unversioned library but check that it is built against the correct version of ngraph

* review comments

* Faster liveness.

Memory manager optimized for non-sharing of tensors.
Add pass manager profiler.

* Move pass profiler to a separate PR

* Move Memory Layout optimizations to a separate PR

* use find instead of count

* Cache some generated functions in backwards tests to speed performance

* more caching

* complete the new backend construction/destruction API
* close each dlopen
* don't close libraries for now as it causes python to segfault

* Usage of mkldnn reshape updated

* update reshape condition for mkldnn

* Add a test case and order in which conditions are checked

* collect matched nodes

* clear m_matched_list

* tests

* address feedback

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

* clean up.

* nbench cleanup

* update style

* 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.

* - 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

* Reshape 4d

* Support dimshuffles/transpose with MKLDNN

* Addressing PR Feedback

* Use Eigen for 3D dimshuffles

* constant broadcast folding

* code review feedback

* 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

* updated to work with llvm 8.1 that tensorflow is built with

* sane extensions on the mac

* not doing rpath on apple

* apply style

* inplace compute

* fix warnings

* address bob's feedback

* bob's feedback 2

* bobs feedback 3

* address bob's feedback 4

* remove reference to ngraph core code from codegen. add stand-alone implementations of needed funcions

* fixed potential pointer leak

* clean up file_util

* more file util cleanup, removing unused functions

* interpreter works on mac

* CPU and INTERPRETER build and pass unmit tests on macos

* move get_directory to file_util

* cleanup