* Register cinn_dialect & Generate AddOp

* codestyle fix

* codestyle fix02

* 合并cinn_ops.parsed.yaml和cinn_backward_ops.parsed.yaml;
将add算子注册进CinnDialect

* 通过宏定义注册算子

* bug-fix

* bug-fix

* bug-fix

* bug-fix

* skip conflict include

* resolve conflict

* remove empty block program

* update implementation

* test=document_fix

* test=document_fix

* support ir api form prim

* convert vector of int to intarray

* support ir api for prim

* support vjp prim mode in new ir

* remove useless code

* add test for prim

* modify utils

* remove useless code

---------
Co-authored-by: Ncyber-pioneer <chenzhuo@tju.edu.cn>

* [Doctest]fix No.198-199, test=docs_preview

* [Doctest]fix No.199, test=docs_preview

* abandon fluid

* fix ()

* fix code

[Inference] Make share_external_data supports bf16 and bool; fix while_op cache_inference_while_scope when using fleet_executor. (#56055)

* 1. make share_external_data supports bf16 and bool; 2. don't drop_kids when cache_inference_while_scope

* fix FLAGS_cache_inference_while_scope

* add unitest

* add unitest

* skip unitest when cudnn_version < 8100

* skip test share_external_data_bf16 when CUDA_ARCH < 80

* fix op translator reshape type

* update

* new ir support vector type place transfer

* add test case

* update

* revert code

* add test assert new ir test

* update

* update

* [Custom Device]add lr_ratio support of adamw for mlu

* fix

* fix

* fix sample codes

* fix bug

* fix bug

* fix bug

* fix shadow output translator

* fix coverage ci

【New IR】 backward gradients accumulate test and pulish append_backward_ops func for op_pattern  (#56265)

* [prim][newir] add basic framework for primitive

* support desctensor in new ir

* add vjp interface

* support vjp in new ir

* support vjp in new ir

* polish vjp interface

* fix stop_gradients set

* fix vjp dispatch

* add comment

* add vjp test for new ir

* add test for tanh vjp

* [prim][newir] add basic framework for primitive

* support desctensor in new ir

* support vjp in new ir

* support vjp in new ir

* polish vjp interface

* fix stop_gradients set

* fix vjp dispatch

* add comment

* add vjp test for new ir

* add test for tanh vjp

* add eager and static backend for warp lower level api

* support call_vjp pybind

* polish code and add test for vjp

* remove useless code

* polish code

* remove useless code

* support mean vjp

* backward origin code

* add test for mean vjp and support has_vjp function

* fix call_vjp

* polish code

* add attrs and dtype interface

* add primitive ops set for backend

* fix compile bugs

* fix some bugs

* fix windows bugs

* add vjp test for tanh_

* fix inference CI

* fix inference ci

* modify fluid cmake

* origin test of tanh and mean passed

* fix conflict

* modify stop_gradient

* remove useless deps

* add cmake

* modify block.ops

* modify test

* fix conflict

* reply review comments

* reply review comments

* pulish code

* fix comment

* fix test

* polish code

* modify backward stop_gradients

* modify static_backend.cc

* refactor grad_op

* support add and add_inplace vjp

* remove useless code

* remove useless code

* remove cout

* modify add_n

* modify add_n with add_vjp test

* modify add_n with add_vjp test

* fix conflict and concat call_vjp

* modify backward test

* Add more gen api

---------
Co-authored-by: Ncxxly <chenxx_id@163.com>
Co-authored-by: NCharles-hit <wanghao107@baidu.com>
Co-authored-by: Nzhangbo9674 <zhangbo54@baidu.com>
Co-authored-by: NYuanRisheng <yuanrisheng@baidu.com>
Co-authored-by: N0x45f <wangzhen45@baidu.com>

* update

* fix codestyle

* update

* update

* add vjp autogen v1.0

* resolve type conflict

* trace run when sync bn

* test=docs_preview

* test=docs_preview

* remove execution_config.used_for_control_flow_op

* update

* update

* open static build flag

* close static build flag

* open static build flag

* add searchsorted to analyze dtype list

* recover and add test_searchsorted_op to static build list

* Update CMakeLists.txt

* Update CMakeLists.txt

* Fix and add unittest

* don't introduce assign when already in global block.

* fix more

* fix bugs

* fix bugs

* fix ci

* fix bfgs

* make function local

This reverts commit fd765f61.

* support setting backward prim blacklist

* add test case

* polish prim forward flag

* fix test case

* [Custom Dice]add run_check support for custom device

* fix error msg

* fix typo

* update for all custom device

* fix

* add warning msg

* [paddle-TRT] support mark output

* [fix bug] hook function only call one in different predictor

* add api test

* fix op translator reshape type

* new ir support vector type place transfer

* add test case

* fix new ir bn grad bug

* add lu_unpack data check

* add error input api test

* add error type info

* add flags for cinn test

* rename DescTensor

* remove useless code

* modify code style

* modify code style

* modify code style

* fix op translator reshape type

* new ir support vector type place transfer

* add test case

* [WIP] Add mp_all_reduce asynchronize overlap.

* Fix some problems.

* Fix dw compute bug, and use a temporary solution to achieve overlap.

* Use fused_linear_param_grad_add to compute dw.

* Reformat ColumnParallel _overlap_linear. Use environment flags to
control following behaviors:
1. export Flags_mp_aysnc_allreduce=True to turn on mp async all_reduce
2. export Flags_skip_mp_c_identity=True to skip two c_identity operators
   in dygraph mode.
3. export Flags_fused_linear_param_grad_add to enable fused_linear_param_grad_add
   in ColumnParallel backward with mp async all_reduce.

* Polish code.

* Remove useless communication API.

* Fix some problems in mp_async_all_reduce and skip_c_identity.

* Add test cases.

* Remove environment variable Flags_fused_linear_param_grad_add in test case.

* Reset error threshold.

* Reset threshold in test case.

* Add useful log. Remove useless test cases.

* [NewIR] support c_broadcast

* add legacyOpList

* refine static op return val