* cherry-pick PR #27933

* fix: cuda version is in varibale CUDA_VERSION in 1.8 cuda.cmake

* close unittest failed temporarily

* cherry-pick PR #27544, fix layer_norm and softmax bug in tensorRT

* fix tensorrt output shape error

* fix unittest tensorrt_engine_op_test

* fix code style for unitest

* cherry-pick a22ea652

* fix analysis_config bug. (#29304)

* fix code format
Co-authored-by: NWilber <jiweibo@baidu.com>

* update.

* fix trt error in ssd model.

* 裁剪transformer模型trt支持；修复tensorRT不支持DeletePass的bug (#28517)

* skip_layernorm_op done

* add unittest

* slice op convertor support trt < 6

* skip_layernorm only work in ernie

* fix unittest

* fix unittest

* Fix TRT plugin registry without TRT lib (#25982)

* fix trt plugin registry without trt lib

* support trt4

* refine code style

* pick ea851796 from develop

* cherry-pick develop PR  #26273 && #27796

* fix unittest error

* fix unittest error

* remove const_cast
Co-authored-by: NPei Yang <peiyang@baidu.com>

* cherry-pick:Add double grad in Squeeze and Unsqueeze to release/1.8, test=develop
Signed-off-by: NTeslaZhao <zhaolisoftware@163.com>

* cherry-pick:Add double grad in Squeeze and Unsqueeze to release/1.8, test=develop

* add matmul doublegrad op

* fix compile errors

* modify code according to review

* delete float16

* delete GetDimForInput to be consitent with release/1.8

[Cherry-Pick] Add Log double grad kernel (#27604)

 double grad support for abs 

* add batch norm double grad, test=release/1.8

* update, test=develop

* fix, test=develop

* optimize slice TRT plugin

This patch removes unnecessary barrier for data transfer of needed offset,
so data transfer can be overlap with GPU kernel execution.

This patch also fixes incorrect name of slice plugin. That is, replaces
"layernorm" with "slice"

test=develop

* add serialize/deserialize to slice plugin

* add static shape slice trt plugin

* fix slice trt op convertor dynamic shape bug

* fix format by clang-format

* fix pylint format error

* fix problems commented by peiyang
Co-authored-by: NRyan Jeng <rjeng@nvidia.com>
Co-authored-by: NShang Zhizhou <shangzhizhou@baidu.com>
Co-authored-by: NRyan Jeng <rjeng@nvidia.com>

test=release/1.8

* Fix test_gast_with_compatibility.py due to the problem of gast in python3.8 (#27433)

test=develop

* fix dll load bug on windows from python3.8 (#27324)

* Support python3.8 (#26850)

* Support python3.8

test=notest

* Replace the 'spawn' start method with 'fork' start method for multiprocessing, on MacOS with python>=3.8 (#27317)

* Replace the 'spawn' start method with 'fork' start method for multiprocessing, on MacOs when python>=3.8

test=develop

* Correct the error in decorator.py (#27409)

test=develop
Co-authored-by: NZhou Wei <52485244+zhouwei25@users.noreply.github.com>

* [Paddle-TRT] Stack op plugin (#25605)

* add stack_op to CMakeLists

* add dim=3 support for scale op

* add trt stack op, test=develop

* remove debug message

* add stack plugin serialize

* remove slice, scale op, will add later

* enhence error message

* revise trt ernie test to conver the stack op CI testi, test=develop

* add stack op serialization

* fix test shape after adding stack op

* remove slice op, will add after implementing serialization

* roll back to min_graph=5 to avoid using slice op

* fix scale op output layer

* implement stack op createPlugin

* use workspace and move the defination to .cu

* move stack plugin creator definition to .cu, test=develop

* sync ut with develop
Co-authored-by: Nzlsh80826 <zlsh80826@gmail.com>

* fix multihead matmul shared params (#27121)

* fix multihead matmul shared params

* add DISTRIBUTED_TRAINER_ENDPOINTS env for cloud

test=develop

* add double grad for expand, test=develop

* add double grad to reduce sum

* delate some wrong message test=develop, test=document_fix

* delate some wrong message

cherry-pick fix cvm check
test=develop
Co-authored-by: N123malin <malin10@baidu.com>

* fix eigen in push sparse; fix hadoop command
test=develop

* add log in load_combine_op
test=develop

This commit fixs the compiling bug regarding unique_ptr of IOptimizationProfile.

IOptimizationProfile has protected dtor and is controlled by TensorRT
internally. Application shouldn't delete the pointer of IOptimizationProfile.
See TensorRT document: https://docs.nvidia.com/deeplearning/sdk/tensorrt-api/c_api/classnvinfer1_1_1_i_builder.html#a9ac47e100454151d8206ac91d543299a
test=develop
Co-authored-by: NJeng Bai-Cheng <jeng1220@users.noreply.github.com>