* Just set threshold to 6 in op mapper relu6

* Remove attrs in op mapper relu6

* onednn: remove fc+eltwiseadd fusion pass
* onednn: remove post-sum fusion in fc kernel
* onednn: tests: make unfused add run into f32

* Fix test failed on cudnn.

* Fix codestyle.

* add split spmd rule

* add pytest in cmake file

* small fix

* fix_ir_printer

* Update ir_printer.cc

* Update ir_printer.cc

* Update ir_printer.cc

* Update ir_printer.cc

* Update ir_printer.cc

* Update paddle/ir/core/ir_printer.cc
Co-authored-by: Nkangguangli <kangguangli@hotmail.com>

* Update ir_printer.cc

---------
Co-authored-by: Nkangguangli <kangguangli@hotmail.com>

* add shard tensor api

* add DistAttr api

* add unittest for coverage

* fix process mesh sample code

* fix checking error

* IntArrayAttributeStorage

* fix new launch

* fix ps uit

* fix empty_var_name problem

* fix coverage ci

* fix coverage ci

* fix cudnn 8.7+ bug on cudnnConvolutionBiasActivationForward

* save_optimized_model_pass support gpu

* add assert

* fix

* fix

* [尝试] 给tensor增加一个属性, 这个属性是一个定值 1

* 暴露gradnode 并构建gradnode新的方法(用来测试)进行暴露给python python端可以访问

* 开发grad_fn、next_functions两个API 并暴露到python端- 做一些规范化处理

* 增加一个单元测试

* 优化 code-style

* 将单侧文件迁到正确的位置

* 优化 code-style

* 删除无用注释

* 解决 __main__ has no attribute

* 修改单侧文件

* 修改单侧脚本-temp

Issue #55016

In preparation for the improvement of the graph compiler, the deprecated old graph compiler was cleaned up.

* add share api for DependencyBuilder

* add judge codes for sharing build results

* add ShareBuildResultsFrom

* update ShareDependencyFrom

* fix error

* add share codes

* fix memory error

* update according review

* update notes

* fix code style

* remove const_cast

* fix code style

* add feed with place op

* remove useless unitest

* udpate mkldnn

* update

* new ir support builtin slice op

* fix phi kernel adaptor bug

* add enable_static

* remove useless test case

* change feed list to single variable

* support gpu

* fix bug

* remove template

* add more data type

* fix cimpile bug

* Fix TRT multihead matmul UT failure

* [cmake] add third party jitify cache

* fix bug

* fixed

* fix bug

* fix hsigmoid_loss

* add test into whitelist

* fix whitelist

* update broadcast gloo in dygraph

* update

* update reduce gloo in dygraph

* update reduce gloo in dygraph

* update

* update allreduce allgather

* update all

* update

* update

* update

[XPU][PHI Kernels] bind reduce_max_int64 set_value_bool sin_grad_fp32 cos_grad_fp32 for XPU (#55375)

* bind kernels for xpu

* format code

* format code

* 0d support for set value

* refine set_value

* [Kunlun] Mofify some legacy code on distributed training

There were limitations on XPUs before, such as concat/split is not
supported, and c_broadcast only support fp32. These limitations are
lifted recently.

Multi-device profiling on XPU will also be supported by this PR.
Without this PR, a hanging broadcast will be issued by devices that
enables profiling, eventually lead to kernel timeout error.

* fix typo