* support 0D for paddle.sort/argsort

* support 0D tensor for paddle.sort/argsort in xpu

* fix bug

* fix grad and add value assertion

* migrate shaple sgd, split,sign xpu kernels to phi, test=kunlun

* fix dlrm throughput problem, test=kunlun

* nullptr bugfix for XPU pg mode

Also a few kernels is added to xpu whitelist

* increase error msg length

* add stat tool

* add roll and roll_grad kernels and strided_slice and strided_slice_grad kernels, test=kunlun

* add xpu rmsprop centered, test=kunlun

* Replace mutable_data with DeviceContext.Alloc in phi kernels

* Fix CI errors

* Fix CI errors

* Fix CI errors, test=kunlun

* Fix CI errors, test=kunlun

* Handle rnn_functor

* Update approvals

* [Eager] Fix paddle.grad interface

* [Eager] Support minimum SubGraph for GeneralGrad

* Add needed_nodes to prune grad graph more thoroughly

* [Eager] Add grad_node_trans_mapping_ to record which grad_node has been transformed to AccumulationNode

* [Eager] Fix paddle.grad interface

* Polish code

* remove potential_stop_node

* Add endding_nodes to enhance genSugraph logic

* clear endding_nodes_

* polish code

* rename endding_nodes to endding_nades_

* Refactor grad interface

* Add register_hook case to fix coverage-ci

* Fix code format

* Refactor general_grad

* Add more code comments

* call clear directly to release GradSlotMeta

* fix a mistake

* fix matmul/ multiply kernel logic and optional input in yaml, fill zeros logic and so on.

* fix batch_norm_double_grad yaml optional config

* fix tanh_triple_grad yaml and kernels

* fix MultiplyTripleGradKernel optional logic

* fix merge mistake

* fix compile error

* remove legacy attr for bn

* polish code

* fix some kernel

* merge develop

* fix error

* remote log

* fix kernel with full like

* hide value log behind

* hide value log behind

* fix matmul_triple grad

* fix xpu compile error

* fix xpu compile error

* fix xpu ut

* fix xpu ut

* fix_xpu_compile_error
Co-authored-by: NWeilong Wu <veyron_wu@163.com>

* [Eager] Fix paddle.grad interface

* [Eager] Support minimum SubGraph for GeneralGrad

* Add needed_nodes to prune grad graph more thoroughly

* [Eager] Add grad_node_trans_mapping_ to record which grad_node has been transformed to AccumulationNode

* [Eager] Fix paddle.grad interface

* Polish code

* remove potential_stop_node

* Add endding_nodes to enhance genSugraph logic

* clear endding_nodes_

* polish code

* rename endding_nodes to endding_nades_

* Refactor grad interface

* Add register_hook case to fix coverage-ci

* Fix code format

* Refactor general_grad

* Add more code comments

* call clear directly to release GradSlotMeta

* fix a mistake

* fix matmul/ multiply kernel logic and optional input in yaml, fill zeros logic and so on.

* fix batch_norm_double_grad yaml optional config

* fix tanh_triple_grad yaml and kernels

* fix MultiplyTripleGradKernel optional logic

* fix merge mistake

* fix compile error

* remove legacy attr for bn

* polish code

* fix some kernel

* merge develop

* fix error

* remote log

* fix kernel with full like

* hide value log behind

* hide value log behind

* fix matmul_triple grad
Co-authored-by: NWeilong Wu <veyron_wu@163.com>

* add stat tool

* add roll and roll_grad kernels and strided_slice and strided_slice_grad kernels, test=kunlun

* add silu, unfold and their grads,test=kunlun

* rename kernel for top_k, slogdeterminant, generate_proposals_v2

* fix bug

* rm fluid “xpu_header.h” deps in phi

* move part of xpu_op_list.h from fluid to phi

* add fluid xpu_op_list deps

* add glog deps for xpu_op_list in phi

* fix PR-CI-Kunlun

* add fluid_op_name_map

* rename some kernel name

* add comments for op-kernel map

* refine map name of op to kernel

* add stat tool

* add roll and roll_grad kernels and strided_slice and strided_slice_grad kernels, test=kunlun

* add masked_selected_grad kernel,test=kunlun

* refine reduce_all

* correct sync behavior for XPU distributed training

XPU support event mechanism similar to cuda event, so it is advisable to
use an event to sync compute/comm streams for performance. However this
mechanism is never fully tested, and inconsistent loss/ending_epochs are
reported. Therefore, this PR replaces event sync with stream waiting as
a temporary solution.

* remove compile warning

* standard api

* fix xpu bugs

* add stat tool

* add roll and roll_grad kernels and strided_slice and strided_slice_grad kernels, test=kunlun

* embedding and embedding_grad add int32 input, test=kunlun