Skip to content

Separate HTTP core into subfolder#33

Closed
letmaik wants to merge 31 commits intoNonStatic2014:masterfrom
letmaik:letmaik/http-core
Closed

Separate HTTP core into subfolder#33
letmaik wants to merge 31 commits intoNonStatic2014:masterfrom
letmaik:letmaik/http-core

Conversation

@letmaik
Copy link

@letmaik letmaik commented Apr 9, 2019

HTTP core has no dependency on protobuf or ONNX RT, only on Boost and re2. This makes it easier to reuse and keeps the architectural structure clean.

Note: This contains the changes from #32 as well, so merge that one first.

Changming Sun and others added 30 commits April 4, 2019 13:03
* Add unique identifier in function subgraph

* enable test
Currently, when using OrtEnableProfiling to enable profiling using the C API,
the profile output file is created but is always empty.

The reason is that InferenceSession::EndProfiling() needs to be called to
write the profiling data to the output file.

However there's currently no way to call this function via the C API.

This adds a call to EndProfiling() to the descructor of the session if
profiling is enabled in the session options.
)

Previously all initializers had to have matching graph inputs and the NodeArg was guaranteed to be created via graph input processing.
* Expose parallel execution option

* delete unnesary file

* add doc

* update nuget retore to 4.3.0

* resolve comments

* remove unnessary file

* make git ignore csharp/Directory.Build.props

* fix yaml config for nuget 4.3
If it is needed, it can be used explicitly in mkldnn provider.
CustomRegistry should use composition instead of inheritence
* Add optimization level as cmd line arguments

* fix the help info and add option.
…ta (microsoft#777)

* Copy required inputs and outputs directly in InferenceSession::SaveModelMetadata

* trivial

* trivial
* Adding a custom op interface to the C API to remove shared library dependency.
* Remove old custom op test
* Rework how custom ops handle inputs/outputs to enable custom op output shape calculation in the compute method
* Add a nicer C++ API for custom ops and switch the tests to use it.
…nsorrt option. we can set it implicitly. (microsoft#780)

* use_full_protobuf if tensorrt build option is enabled.

* update BUILD.md sections on MKLDNN and TensorRT/full_protobuf option
@tmccrmck
Copy link
Collaborator

tmccrmck commented Apr 9, 2019

Hey Maik, could you rebase against master after #32 is merged? I'll merge it as soon as the build completes.

Edit: thanks for doing this. I was actually going to do this today :)

Edit 2: I can also do it. I know it's late in the UK.

@letmaik
Copy link
Author

letmaik commented Apr 9, 2019

No worries, I've rebased (well, merged) just now. Good that you had the same idea! :)

@letmaik letmaik closed this Apr 9, 2019
@letmaik
Copy link
Author

letmaik commented Apr 9, 2019

See #34. CI doesn't run on forks.

NonStatic2014 pushed a commit that referenced this pull request May 14, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.