GitLab

Merge branch 'dyn_ref_multibroadcast' of github.com:ROCmSoftwarePlatform/AMDMIGraphX into dyn_transpose

* Update migraphx docker container to include half and cmake 3.24.2

* Include timestamp while tracing

Updated Multibroadcast op to have a two input version for dynamic shapes
Current dynamic shape broadcasting logic
dynamic_dimensions must be the same or one of them is {1, 1, 0} or {1, 1, 1}
Works for dyn-dyn, dyn-static, and static-static shape combinations
Changed common.cpp for multibroadcasting for binary ops with dynamic shapes
Extended binary.hpp for dynamic shapes to test the new common.cpp stuff

Co-authored-by: kahmed10 <15948690+kahmed10@users.noreply.github.com>

* free up more github runner space
* upgrade versions

Two input version of the broadcast operator to handle dynamic shapes
Added comments to describe the versions of the broadcast operator
Dynamic broadcast only handles broadcasting a static 1D shape tensor into the other input shape

Can be enabled via environment variable MIGRAPHX_ENABLE_NHWC

Allows for a model to be converted to the same opset but turn on infer_shapes through
onnx. This allows us to get an idea of what should be valid for the output of nodes
in a network.

Usecase:

python3 tools/convert_onnx_version.py --model <model_name> --opset=<same_as_model> --infer_shapes --output <new_model_name>

Trying to keep the PRs separate

Merge branch 'dyn_ref_multibroadcast' of github.com:ROCmSoftwarePlatform/AMDMIGraphX into dyn_broadcast

Newer split moves the split attribute to an input. In this case we check the
number of input args then.

Merge branch 'dyn_ref_multibroadcast' of github.com:ROCmSoftwarePlatform/AMDMIGraphX into dyn_transpose

Merge branch 'dyn_ref_multibroadcast' of github.com:ROCmSoftwarePlatform/AMDMIGraphX into dyn_transpose

Merge branch 'dyn_ref_multibroadcast' of github.com:ROCmSoftwarePlatform/AMDMIGraphX into dyn_broadcast