This reverts commit 20e68a22.

Co-authored-by: Ntianshuo78520a <707759223@qq.com>

* add dla

* add python api
Co-authored-by: Nshangzhizhou <root@szth-rp-fanyi-opera49.szth.baidu.com>
Co-authored-by: Nshangzhizhou <root@szth-rp-fanyi-opera49.szth.baidu.com>

* fix datanorm error msg (#30294)

* Optimize the error message of framework. (#30134)

* modify error message based on comments (#30189)

* modify error message based on comments

* edit code according to review.

* Correct spelling according to review.

* fix enforce msg of sum xpu op (#30113)

* enhance error info for py_func (#30138)

* enhance error info for py_func

* update

* fix elugradgrad test fail & error message opt (#30171)

* fix elugradgrad test fail and error message opt

* fix unitest,test=develop

* Update prroi_pool_op.h

fix error message

* opt message,test=develop

* fix ci fail,test=develop

* Refine PADDLE_ENFORCE Error Messages. test=develop (#30149)

Improve some error messages in parallel_executor.cc, conditional_block_op.cc, recurrent_op.cc

* enhance error message, test=develop (#30220)

* fix error message for distribute_fpn_proposals_op (#30116)

* enhance error msgs of fusion_seqpool_cvm_concat_op.cc, test=develop (#30240)

* just add the op error message for the matmul xpu (#30246)

 add the op error message for the matmul xpu

* enhance error message of nll_loss op test=develop (#30125)

* enhance error message of nll_loss op test=develop
Co-authored-by: Nyaoxuefeng <yaoxuefeng@baidu.com>
Co-authored-by: Nxiemoyuan <71377852+xiemoyuan@users.noreply.github.com>
Co-authored-by: NWeiXin <weixin10@baidu.com>
Co-authored-by: NJack Zhou <zhoushunjie@baidu.com>
Co-authored-by: NWilber <jiweibo@baidu.com>
Co-authored-by: NDouble_V <liuvv0203@163.com>
Co-authored-by: NHuihuang Zheng <zhhsplendid@gmail.com>
Co-authored-by: Nzhang wenhui <frankwhzhang@126.com>
Co-authored-by: Nwangguanzhong <jerrywgz@126.com>
Co-authored-by: N石晓伟 <39303645+Shixiaowei02@users.noreply.github.com>
Co-authored-by: Nwawltor <fangzeyang0904@hotmail.com>
Co-authored-by: Nlijianshe02 <48898730+lijianshe02@users.noreply.github.com>

* remove -DSUPPORTS_CUDA_FP16 in cuda.cmake

* comile with cuda9

* add some unittest

* notest;test=coverage

* add unittest for trt plugin swish && split

* update ernie unittest

* fix some error message

* remove repeated judgement of CUDA version in mbEltwiseLayerNormOpConverter

* fix comile errror when CUDA_ARCH_NAME < Pascal"

* fix comile error

* update unittest timeout

* compile with cuda9

* update error msg

* fix code style

* add some comments

* add define IF_CUDA_ARCH_SUPPORT_FP16

* rename IF_CUDA_ARCH_SUPPORT_FP16 to CUDA_ARCH_FP16_SUPPORTED

* change avg pooling and global pooling to trt layer

* add support for static shape global pooling

* modify trt errmsg

* skip_layernorm_op done

* add unittest

* slice op convertor support trt < 6

* skip_layernorm only work in ernie

* fp16 result ok

* change -DWITH_NVINFER_PLUGIN toconfig.EnableTensorRtOSS

* auto detect special slice op converter for ernie with trt oss

* ernie oss only support fp16

* fix special_slice_plugin serialize bug

* matmul in tensorrt ok

* ernie unittest ok

* add matmul tensorrt unittest

* remove demo code

* add info log for trt input dynamic shape check

* fix error msg error

* add unittests and op version register for tensorrt_subgraph_pass

* rename to test_trt_subgraph_pass.py

* fix softmax converter diff when padding dim=1

* use iwyu clean include, test=develop, test=win

* compilation error, test=develop

* fix compilation error2, test=develop

* fix compilation error3, test=develop

* fix compilation error4, test=develop

* fix compilation error5, test=develop

* fix compilation error6, test=develop

* fix compilation error7, test=develop

* fix compilation error8, test=develop

* fix compilation error8, test=develop

* fix compilation error10, test=develop

* fix compilation error11, test=develop

* polish some lost error msg

* add some math file to white list

* polish detail based reviewer commnet

* 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>

* optimize errror report

* add test case for pad op converter

* fix some spelling mistake commented by peiyang

test=develop

* 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

* support trt dynamic shape int8

* add unittest

* add support for sigmoid; adapt to trt6+ api

test=develop

* fix trt plugin registry without trt lib

* support trt4

* refine code style

* develop dynamic shape serilization

* add test param for gelu

* fix bugs

* delete redundant comments

* debug

* fix conflict. test=develop

* fix bug. test=develop

* add trt dynamic shape serialized support

* fix ernie serialized bug
test=develop

* fix codestyle
test=develop

* fix bug
test=develop

* fix bug.test=develop

* modify cmakelist test=develop

* fix bug
test=develop

* fix error message.  test=develop

* fix trt register plugin based on pr#25003

* add trt dynload

* fix deserialization bug of not finding plugin registration

* refine code style

* recover engine key in tensorrt_subgraph_pass

* for ci coverage

* add unittest for deserialization
Co-authored-by: Nhaozech <chenhaoze94@gmail.com>

Use vector instruction (LDG.128) to improve qkv transpose. It
provides 1.4X speedup at same GPU base frequency.
test=develop

* fix multhead matmul's instable
test=develop

* fix multihead matmul bug
test=develop

* fix converage problem
test=develop