onnxruntime/onnxruntime/core/graph
Wei-Sheng Chin de9da123cf
Enable static memory planning for pipeline. (#4204)
* Enable static memory planning for pipeline.
1. We fix a bug when resolving symbolic shape for scalars.
2. We pass the original inputs to all pipeline stages so that
   the symbolic shapes can be resolved.

* Further Improvements
1. Address comments.
2. Further reduce activation size by ~50% when pipeline is on.
   This is done by removing all but one gradient tensor from the last
   RecordEvent in the backward pass.

* Address a comment

* Fix Windows build
2020-06-12 21:43:50 -07:00
..
contrib_ops matmul integer fusion (#4195) 2020-06-11 21:42:09 -07:00
dml_ops Update onnx submodule to 1.7.0 release candidate (#3405) 2020-04-04 16:23:42 -07:00
featurizers_ops Changes on RollingWindow Transformer (#3679) 2020-04-24 13:00:49 -07:00
function.cc Function expansion support and Update ONNX to 1.7 release candidate 1. (#3782) 2020-05-01 10:35:16 -07:00
function.h Initial bootstrap commit. 2018-11-19 16:48:22 -08:00
function_container.h Document the Graph header files and cleanup some issues. (#42) 2018-11-28 08:42:11 -08:00
function_impl.h Avoid using the default logger in the graph lib and optimizers (#2361) 2019-11-14 13:23:28 -08:00
function_inliner.h remove graph editor since it's not designed as expected to restrict graph access for rewrite rule. (#119) 2018-12-06 14:10:51 -08:00
graph.cc Enable static memory planning for pipeline. (#4204) 2020-06-12 21:43:50 -07:00
graph_utils.cc matmul integer fusion (#4195) 2020-06-11 21:42:09 -07:00
graph_utils.h matmul integer fusion (#4195) 2020-06-11 21:42:09 -07:00
graph_viewer.cc Implementation of Nuphar execution provider (#881) 2019-09-01 23:01:47 -07:00
model.cc Enlarge the read buffer size in C#/Java test code (#4150) 2020-06-08 16:13:11 -07:00
model.h resolve conflict 2020-03-20 20:20:35 +00:00
op.cc Bump onnx to latest (#1756) 2019-09-12 11:46:49 -07:00
op.h Avoid using the default logger in the graph lib and optimizers (#2361) 2019-11-14 13:23:28 -08:00
record.h Make CentOS 6 CUDA build and run (#2159) 2019-10-19 15:33:31 -07:00
schema_registry.cc Allow a custom op with the same name to be registered for several providers. (#3400) 2020-04-02 15:38:51 -07:00