* transfer as_real to phi

* fix erros

* blocking: True -> False

* eager variable back sync

* update xccl lib
    * use separate streams for compute/comm on XPU
    * add broadcast op to xpu2_op_list

* add fused token prune op and plugin

* sparse support amp

* EagerAmpAutoCasts support sparse

* build into one static library

* move memory/detail to memory/allocation

* fix bug

* fix profiler

* fix framework_proto

* fix deps

* fix inference compilation

* fix rocm compile

* follow comments

* fix buddy_allocator_test

* Compilation optimization

* adapt mkldnn kernel in PHI

* fix ci compile bugs

* fix compile bugs

* fix compile bugs

* fix compile bugs

* fix compile bugs

* delete comment

* fix compile bugs in windows-inference

* delete code for converage

* modify code by review

* modify code by review

* add todo

* fix compile bugs

* fix compile bugs

* fix compile bugs

* fix unittest bugsx

* [CustomDevice] add custom ccl api

* add ut

* add amp logic in python_C

* fix inplace bug

* Move eigvals OP to PHI

* Fix CI errors

* Fix CI errors

* [Phi] Migrate matrix_solve to phi

* replace mutable_data with Alloc

* mv to phi

* refine infermeta code position

* refine grad code

* add api yaml and add final_state_api

* refine code

* [JIT]Layer support eager dygraph mode and polish Function interface

* remove usless code

* fix #define

* add relu6 and relu6_grad

* change code style of relu6 and relu6_grad

* add relu6 and relu6_grad *test=kunlun

* add relu6 and relu6_grad *test=kunlun

add shape attribute in fill_constant op converted from scale_loss_grad after convert graph to program (#43898)

* fix grad loss shape

* little change

* delete for_test

* add unittest for FLAGS_CONVERT_GRAPH_TO_PROGRAM

* avoid conflict

* fix cpu lars_momentum bug & add xpu grad_add/log_softmax/log_softmax_grad
*test=kunlun

* minor
*test=kunlun

* fix convtranspose and elementwise

* grid sampler op for xpu. test=kunlun

* update relu xdnn api. test=kunlun.