* feat: static check

* refactor: replace Collective & PointToPoint with NCCLEnv

* refactor: rename to RunFnInNCCLEnv

* refactor: pass std::function by value

* 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

* fix device id issue for xpu eager

xpu device id is not correctly set in eager mode, thus vars are on dev0 unless
XPUDeviceGurad is called, leading to this error message for all node rank != 0:
"NotImplementedError: (Unimplemented) Place Place(xpu:0) is not supported."

* fix typo

* fix pybind error

* refactor: update pg custom

* fix: use new api in ut

* fix: typo

* revert: recover legacy apis

* fix: add GetDeviceContext

* refactor: simplify send, recv interfaces

* refactor: rm send_partial, recv_partial, all_gather_partial

* XPU support eager mode

* add unittest for XPU eager mode

* minor bugfix

* minor bugfix, test=kunlun

* correct copyright info

* 1. remove unsed vars/funcs
2. ProcessGroupBKCL inherit from ProcessGroupStream

* bugfix for fp16 in eager mode multi-card, test=kunlun

* rebase & fix a few issues

* use new processgroup interface, test=kunlun

* fix compile issue, test=kunlun

* refactor: send, recv, send_partial, recv_partial

* refactor: rm useless const ref

* removing dependent to fluid/framework/eigen.h in phi

* more fix according to PR-CI-Py3 fail

* remove unused-variable warning in linux

* fix unused-variable error in GpuPS

* [Auto Parallel] Improve the c++ dist attr

* [Auto Parallel] Modify test_program.py

* Support custom stream for standalone executor
Co-authored-by: NYulong Ao <aoyulong@baidu.com>

* [CustomDevice] GetCCLComm add custom device support

* update

* update

* update

* [Auto Parallel] Improve the c++ dist attr

* [Auto Parallel] Modify test_program.py

* [Auto Parallel] Add the missiong import

* fix no sync bugs

* update

* update task chain

fix: update wait chain

feat: add `GetDeviceContext` for gloo

* fix oom

* fix dev

* update

* update
Co-authored-by: NLiYuRio <liyuruijx@163.com>
Co-authored-by: NForFishes <2282912238@qq.com>

* make all cpp tests dynamic linked to libpaddle.so

* add comments

* keep old cc_test for some tests

* fix some ut

* make some ut use cc_test_old

* fix typos and fit for win32

* fix lib path

* fix some tests

* skip lite test

* fit for rocm

* fit for cinn

* fit for mac

* fit for win32

* skip inference ut

* skip  windows

* fix coverage

* Close Wno-error of sign-compare

* fix sign-compare in graph_node_test.cc

* fix sign-compare in graph_node_test.cc file

* fix sign-compare in common_graph_table.cc file

* Fix bug of reduce_sum op. When input.numel() > INT32_MAX, its result
is wrong.

* support pure bfloat16

* support bf16 linear

* update PR to pass CI

* tiny fix where_grad_kernel.cu

* Support bfloat16 type for reducer and sharding.

* Fix some bug.

* Polish code.

* Polise code.

* Add bfloat16 datatype in fill_grad kernels.
Co-authored-by: Nsneaxiy <sneaxiy@126.com>

* add rpc module in cpp side

* add rpc module in python side

* support win32 and mac for rpc

* 代码优化

* 优化代码

* update rpc

* update rpc launch

* rpc remove rank and world_size api

* fix logger import bug

* remove support for win and mac

* remove support for xpu, npu, cinn and rocm

* remove support for xpu, npu, cinn and rocm

* fix shutdown barrier timeout bug

* update:python_rpc_handler to shared ptr

* fix master shutodwn first bug

* tests support for cpu

* update log to vlog

* update get service info api

* add single process test case

* remove process group

* remove some useless dependencies

* update rpc api comments

* update rpc comments: Example to Examples

* update rpc api comments

* update rpc api comments

* update launch api comments

* update init_rpc comments

* update rpc sync and async comments

* fix bug: init_rpc cant be called repeatly in a process

* update rpc api comment: make master endpoint unique

* update rpc api:service to worker, timeout_ms to timeout

* rename ServiceInfo to WorkerInfo

* refactor: rename server to worker, log to vlog

* add launch test

* remove unused codes

* refine

* remove using lodtensor part1

* polish history code format