* comm_context and static init

* refactor: move to phi/core/distributed

* refactor: avoid mutable_data usage

* fix: windows sock

* fix: device without nccl
Co-authored-by: Wen Sun <syl1887415157@126.com>

* Enable standalone executor for fleet training

* Update code

* Replace use_standalone_executor utils in auto parallel

* Update code

* Diable standalone executor for test_pass_sharding

* Update code

* Set sequential run for auto parallel

* Fix dist_attr bug

* Set sequential run for auto parallel

* add 0-d support for paddle.kthvalue

* add 0-d support for paddle.mode

* fix coverage test for device

* fix check-bug in windows

* change axis check from LT to LE

* add shape & value check for grad when input is 0d tensor

* zero-tensor

* remove unused

* zero_dim_xpu

* relocate

* add value test

* fix syntax

* [D2SCinn]Add test_cinn unittest and param_grad into skip_gc_vars

* remove print

* acquire custom_deivce stream

* regulate file name and unittest

* flatten op support 0D-tensor

* add test in zero dim py

* fix shape should be list

* short code for ci-coverage

* add backward test

* simple code for ci coverage

* add axis check

* add 0D-tensor test in test_flatten_contiguous_range_op.py

* add axis error test for Coverage CI

* add more test for CI-Coverage

* add more test for CI-Coverage

* [Auto Parallel] Rename methods of ProcessMesh

* [Auto Parallel] Impl the python process_mesh by the c++ one

* [Auto Parallel] Add some minor modifications

* [Auto Parallel] Rename some methods

* [Auto Parallel] Remove unnecessary codes

* [Auto Parallel] Add back some removed files

* [Auto Parallel] Fix bugs

* [Auto Parallel] Fix a bug

* Update process_mesh.cc

* [Auto Parallel] Merge dist attrs of Python into C++

* [Auto Parallel] Add back deleted importing

* [Auto Parallel] Add back removed unittest

* [Auto Parallel] Remove type qualifiers of return types

* [Auto Parallel] Fix some bugs

* [Auto Parallel] Fix a bug of the quant pass

* [Auto Parallel] Fix the code style

sequence_mask fix: when the input length is an empty tensor, the kernel tries to dereference illegal sentinel iterator (#49525)

* support 0D for paddle.sort/argsort

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

* fix bug

* fix grad and add value assertion

* generate the static graph code of ops

* modify the isclose comment

* modify the clip comment in nn.py

* reset nn.py

* update nearest_interp, expand_v2, cast for stable diffusion

* update nearest_interp, expand_v2, cast for stable diffusion

* correct shape rank

* Update expand_v2_op.cc

* add pool flops

* add annotations and tests

* rm op_function_generator

* rm op_func_generator.h

* rm op_function

* modify cmake

* rm op_function.h

* rm check for op_function_generator.cc

* reset imperative

* rm python part

* fix imperative

* lint

* lint

* modify legacy_c

* review

* modify

* modify legacy

* rm gen op_functions code

* reset framework

* rm core.ops for test

* core.ops->core.eager.ops.legacy

* not raiseError for xpu

* [Eager] optimize same python api logic

* optimize full api

* optimize logic

* optimize logic

Add 0d Tensor Test Cases for cond, case, switch_case. Since the 3 APIs are control flow APIs, their support for 0d tensor relies on the underneath APIs. This PR just added test cases to prove that the 3 APIs have already handled 0d tensor well.

* remove deps and prior comm

* grad comm fuse

* add deps for amp&global norm

* stage2 broadcast prior deps

* stage2 grad overlap

* stream_analyzer bugfix

* overlap enable

* dep op namescope

* depend support multiple inputs

* check finite deps

* stage2 param comm overlap

* Set kD2HStream

* grad comm hierarchical

* grad comm hierarchical

* new unitest
Co-authored-by: Nchenruibiao <chenruibiao@baidu.com>

* update requirements.txt and setup.py.in

* update requirements.txt setup.py.in setup.py

This reverts commit 2c444dfa.

* add for-else

* add * for unpacking