* optimize block config  and fp16 atomicAdd perf for lookup_table_v2_grad.

* skip kernel declare for cuda only kernel on rocm

* fix error

* add processgroup_nccl

* runtime supports pten kernels, test=develop

* fixes a bug, test=develop

* move trunc_op's infere shape

* modify according to risheng's comment

* move randperm to phi

* fix npu

* fix memory::Copy

* polish default sig impl

* revert dispenable out

* [MLU] add cncl parallel context and mlu resource pool

* [MLU] fix the cncl_context_test

* change CUDA implementaion of bernoulli OP

* fix CI

*test=kunlun

* move unbind to phi

* revert infer shape

* add header file

* move concat_and_split to phi

* [KP] Add elementwise add xpu, test=develop

* modify the File Permissions

* modify the copyright time

* modify code style

* modify code style

* [Phi] Migrate lable_smooth_op into Phi

* fix PT->PD

* update inference part

* restore white space

* update gather_nd trt converter ut

* update ut

* fix 'is with a literal'

* fix typo

* eager, test=develop

* fix bug, test=develop

* eager, test=develop

* merge legacy to fluid

* eager, test=develop

* eager, test=develop

* Refactor TensorAdd func by template and remove gradient_accumulation in eager

* Remove needless target name

* eager, test=develop

* eager, test=develop

* Use overload instead of template

* Remove legacy code

* Remove legacy code

* selectedrows, test=develop

* Remove DataType test

* eager, test=develop

* eager, test=develop

* support gan, test=develop

* Using Tensor directly instead of using EagerTensor

* support gradient_accumulation

* make test_imperative_lod_tensor_to_selected_rows longer

* make test_imperative_lod_tensor_to_selected_rows longer

* refine code

* ptb, test=develop

* Rename all EagerTensor to Tensor

* Rename some EagerTensor to Tensor

* rename EagerTensor to EagerVariable

* eager, test=develop

* eager, test=develop

* eager, test=develop

* eager, test=develop

* add more test

* eager, test=develop

* Support copiable selected rows and merge develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* clear grad, test=develop

* merge, develop

* merge, develop
Co-authored-by: NJiabinYang <360788950@qq.com>
Co-authored-by: NWeilong Wu <veyron_wu@163.com>

* add elementwise_div

* refine rocm

* refine code

* refine op register

* solve conflict

* refine unittest

* refine unittest precision

* add rocm

* fix RecordEvent interface

* modify default level to 4

* update interface use

* add const default trace level

* update record event interface using

* update record event interface using

* update record event interface using

* update operator.cc

* update part2

* update part1

* update part3

* fix include profiler.h header in ps server

* fix include profiler.h header in ps server

* fix profiler.h header

* fix profiler.h header

* fix merge buf

* update

* fix bug

* fix bug

* Supported intermediate outputs for eager final state codegen

* Added validation check for intermediate tensors

* remove fill_any_like kernel in fluid and fix data transform bug

* support scalar in infershpe

* recover infershape in fill_and_like

* [custom kernel]Delete useless

* change RegType enum names

* mod notes

* merge

* update

* add subblock logic for context and partitioner

* partitioner support sub blocks

* revise typos

* fixed param init bug for while

* chmod 644

* add unitest

* mv forward parser

* update unitest

* update dist op ctx

* update dist op ctx

* fixed bug in dist op ctx

* fixed bug for recompute subblock

* Move Real OP to phi

* Move Imag OP to phi

* Move Real and Imag InferShape to phi

* Move Real and Imag to complex_kernel

* Change PT_REGISTER_XXX to PD_REGISTER_XXX

* add tcp_socket and tcp_store

* delete gather_ut skip_case

* add trt version limit

* Adapt to batch_norm_grad op and add align function in
roi_align op for kunlun, *test=kunlun

* Adapt to batch_norm, batch_norm_grad op api for kunlun, and add unit-tests of batch_norm, roi_align. *test=kunlun

* change Vector to std::vector and provide MixVector class as a helper wrapper class

* solve the multi-gpu hang problem

* remove the duplicate template instantialize

* Copy vector to cpu

* add CopyToCPU

* xxx

* final version: fix the problem of all reduce

* remove mixvector dependence

* fix

* merge

* fix code

* fix by CI