Yinghai Lu
45e5c17ecf
ONNXIFI transform ( #9569 )
...
Summary:
Cut-off runnable subgraph and off-load to ONNXIFI backend
Pull Request resolved: https://github.com/pytorch/pytorch/pull/9569
Reviewed By: Maratyszcza
Differential Revision: D8930408
Pulled By: yinghai
fbshipit-source-id: 2b494f7f8dc10c00e58cf0fed5c4a9434be6155b
2018-07-20 15:09:59 -07:00
Pooya Davoodi
56fa6ec66a
[caffe2] Change iteritems in trt/transform.py to items for python3 compatibility ( #7599 )
2018-05-15 20:32:06 -07:00
Yinghai Lu
9ed46c615c
[Caffe2] Provide option to initialize the TensorRT engine at Operator constructor time ( #6809 )
...
* Try to have a lazy conversion of onnx-trt
* .
* Make it work
* comments
2018-04-23 13:09:35 -07:00
Yinghai Lu
7f587de4bc
[Caffe2] Let TensorRT flow use the generic graph transformer ( #6696 )
...
* Refine the transform API
* Let TensorRT flow use the generic graph transformer
* Rebase
2018-04-19 10:07:01 -07:00
Yinghai Lu
6252706feb
[Caffe2] Workspace centric API for TensorRT transformation ( #6678 )
...
* Workspace centric API for trt transformation
* Merge SSA rewrite code
2018-04-17 21:23:27 -07:00
Yinghai Lu
434f710f3f
[Caffe2] Add support to TensorRT ( #6150 )
...
* Add support to TensorRT
* Removed License header
* Bind input/output by position
* Comments
* More comments
* Add benchmark
* Add warning for performance degradation on large batch
* Address comments
* comments
2018-04-11 17:03:54 -07:00
