* add new profiler components

* fix bug

* upgrade new profiler

* fix operator.cc

* fix operator.cc

* fix cmakelists.txt

* fix bug

* fix according to pr

* fix bug

* fix cmake

* fix bug

* fix a bug

* fix bug

* fix bug

[fleet_executor] Add entrance of FleetExecutor in AnalysisPredictor for distributed inference (#39992)

* unify complex type trait and fix real imag bug

* add unittest for type tratis

* [MLU] adapt matmul op

* [MLU] fix phi namespace

* change CUDA implementaion of randint OP,move distribution common func to phi

* fix CI

* fix CI

* open eager when WITH_PYTHON, test=develop

* refine, test=develop

* refine, test=develop

* add DWITH_PYTHON for gen_fluid_lib, test=develop

* move sequence2batch

* move lstm and gru

* Add phi/kernels directory into exclusion to stop using hipcc to compile non .cu files in it.

* update pd_2_trt lower pass

* update pd_2_trt lower pass

* update style

* udpate

* change trt.graph to trt.create_engine

* update comments

* update comments

* add test

* fix bug of paddle.to_tensor and paddle.moveaxis

* fix CI

* modify infershape utils and rm reduce infershape

* merge develop

* fix infermete bug

* add IsForInferShape func in ArgumentMappingContext

* add reduce_mean infermeta

* modify annotation

* add default dims

* tranfer the selu_op and pass the CI

* add sig files

* fix code

* fix by code review

* remove TOOD

* change the include position

* change the head position

* Add function description for Kernel Primitive API
1. Set cumsum and sort share memory size = 1024
2.sort and cumsum api limitation : blockDim.x must be less than 512 (blockDim.x <= 512)

* Refactored GradNodeAccumulation data structure and behaviour

* Fixed CI issues

* Fix compilation issues

* Fixed minor issues

* Reverted changes for intermediate and OverwriteOutput

* fixed minor issue

* Fixed auto codegen for intermediate tensors

* Removed restriction on AccumulationNode modification

* Fixed CI Coverage issues

* Adjusted Log contents

* Fixed CI issues

* Add mobilenetv3_large performance test

* Disable the BF16 test if the device does not support BF16 computations

* Change test timeout

* add layer norm

* add p norm

* add reduce sum

* refine layer norm register bf16 for cudnn811

* add bf16 cast for hip

* add unittest

* refine rocm

* refine layer_norm unittest

* refine reduce op

* refine unittest

* enhance atol for reduce unittest

* remove

* pass

* more pass

* add scale gather sum

* refine CUDA_ATOMIC_WRAPPER ADD for bf16

* add gather unittest

* solve conflict

* add scale uinttest

* add sum unittest

* solve conflict

* refine gather unittest

* refine unittest

* remove SetAllocationForOutputTenosr

* add place param for copy kernel

* recover SetAllocationForOutputTenosr

* polish code

* fix empty_dev api bug

* remove reseting dtype and layout for output in executor

* fix merge bug

* [Phi] Add ClearHolder when re-alloc on new place in DeviceContext

* fix hostAlloc

* remove setting output allocation

* remove full_kernel_impl.h

* fix bug of xpu full_like
Co-authored-by: NAurelius84 <zhangliujie@baidu.com>

* move uniform_random to phi

* fit selected_rows

* replace mutable_data

* support kps backend and compile

* resolve conflict

* fix kps backend trans

* test in xpu2 device

* remove dummy kernel

* optimize mergeadd for sparse_adam,*test=kunlun

* optimize mergeadd for sparse_adam,*test=kunlun

* optimize mergeadd for sparse_adam, *test=kunlun