* revert p2p communication for xpu

* pp for npu

* update

* update

* fix xpuplace

* add ut for sync send

* Revert "fix xpuplace"

This reverts commit f89c1d7622426686bc153a3414a42c39e0f4a647.

* add ut for pp sync send

* rm unusable ut

* update

To make it synchronized at the first recv operator.
If warping all send and recv operators with group start and end, the received tensor will be not complete.

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* polish

* fix bug: wrong match between depend and c_allreduce_sum

* fix codestyle

* fix bug

* add c_sync_calc_stream back

* fix

* revert

* use flag to control

* fix for code coverage

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* test=allcase

* fix doc erros, test=allcase

* Revert "remove ASCEND* keyword (#53046)"

This reverts commit 7fa415ca.

* Delete ascend_trigger_op.cc

* revert-53046-remove_ASCEND_keyword

* update

* update

* remove ASCEND* keyword

* update docstring

* bug fixed

* bug fixed

* [CustomDevice] add recompute support

* update

* fix

* fix

* fix

* fix

* fix

* fix fuse group order

* rename _varbase_creator as create_tensor

* rename _varbase_creator as create_tensor

* cherry-pick fleet executor from 2.4

* fix test case

* remove hccl in .py files

* remove ascend in setup.py.in

* remove ascend in setup.py

1. modify set_value op, use Scalars to represent attr `values`, instead of a bunch of attributs of various types; (#52408)

2. add program converter and set_value op as an example, which provides the functionality to convert `paddle::framework::ProgramDesc` between old and new formats(the differences are mainly some operators with incompatible updates in the definition);
3. program version and operator version map now are always saved when serializing `paddle::framework::ProgramDesc` to identify the version;
3. provide an option `legacy_format=false` in  serialization of `paddle::framework::ProgramDesc`, it decided whether to convert ProgramDesc back to a legacy format, which is compatible for paddle 2.4.2 or earlier versions to load and execute;
4. deserialization of `paddle::framework::ProgramDesc` is now automatically detecting whether the bytes it receives is in legacy format(contains any of the operators that has been incompatibly updated and have any attribute of type `Scalar`) and convert it to new format. But if you want a faithful deserialization without the automatic conversion, you can use protobuf's deserialization instead. Though it is not recommended, it can be used for the purpose of testing.

* add auto parallel tuner options in launch

* add ut for launch in auto_parallel tuner

fix code format

* fix ci-converage

* remove c_comm_init_hccl_op.cc and c_gen_hccl_id_op.cc

* remove gen_hccl_id_op.cc

* add seed control

* fix bug

* fix mkdir

* update

* rem is_compiled_with_npu

* rem nup related code

* make lint happy

* rem test

* remove some tests

* Update grad_scaler.py

* fix an error

* relocate debugger.py

* fix bug

* fix bug

* fix bug

* fix bug

* rem is_compiled_with_mlu

* fix some mlu_place and mlu_device_coount

* make lint happy

* autofix
Co-authored-by: NLiyulingyue <83450930+Liyulingyue@users.noreply.github.com>

* revert changes in python/paddle/distributed/fleet/utils/hybrid_parallel_util.py

* empty commit, trigger ci

* fix test_slice

---------
Co-authored-by: NSigureMo <sigure.qaq@gmail.com>

* Support ignore_index for c_softmax_with_cross_entropy_op.

* Polish code. Remove useless comments and add Testcase.

* Polish code for TestCase.

* Polish code.

* Polish code style.

* Polish code.

* Change loss calculation formula and ignore_index dtype.

* Polish TestCase.

* Fix bug of c_softmax_with_cross_entropy_op_xpu_op. Attribute 'ignore_index'
dtype is int64_t.