* [Eager] fix max_pool3d_with_index interface under different mode

* fix mistake

* Add tests under legacy and supplement comments

* bugfix (#45332)

* customize wait_comm

* add dist cuda kernel

* reuse some funcs in phi

* 使用pnorm

* fix code style - explicit

* fix code sytle

* fix bug

* remove unused headers

* add batch norm xpu kernel

* add batch_orm_grad xpu kernel, test=kunlun

* fix lost changing error, test=kunlun

* remove batch norm op xpu, test=kunlun

* fix dtype error, test=kunlun

* fix const deduce error, test=kunlun

* [XPU]Migrate Adam XPU kernel into Phi

* test=kunlun

* clean includes of convert_util, test=kunlun

* add deps on mkldnn, test=kunlun

* make dense_contents_ lazy init

* update legacy dygraph

* fix legacy dygraph bug

* [NPU] add conv3d_transpose, instance_norm, instance_norm_grad

* add ut

* remove instance_norm_grad

* add erf_p primitive operators

* add gelu orig2prim rule

test=kunlun

* copy kernel file to phi

* delete some code

* migrate uniform_random, test=kunlun

* fix input error, test=kunlun

* fix gpu register error, test=kunlun

* add include file, test=kunlun

* try fix error from CI, test=kunlun

* polish other PR

* fix CI-coverage error, test=kunlun

* add fetch and prune for build cinn pass

* add prune flag

* tranfer xpu: conv2d into phi

* tranfer xpu: conv2d into phi
test=kunlun

* test=kunlun

* test=kunlun

* test=kunlun

* [XPU]Migrate adamw XPU kernel into Phi

* test=kunlun

* test=kunlun

* config

* fix unittest

* zero init & cache & patch config

* add barrier to save and load

* add unittest

* migrate slice/slice_grad to phi, test=kunlun

* minor, test=kunlun

* typo, test=kunlun

migrate xpu activation/activation_grad/transpose/transpose_grad/tril_triu/tril_triu_grad kernel to PHI, test=kunlun (#45554)

* migrate coalesce_tensor_xpu, test=kunlun

* fix xpu SetConstant, test=kunlun

* fix include error, test=kunlun

* refine cmake of framework

* add deps for dense tensor

* fix deps

* remove alloc(ctx)

* add depends on mkldnn

* enhance grid_sampler cpu kernel to 5D input

* fix bug when 5D input tensor running on the cudnn kernel

* move elementwise_floordiv、elementwise_max、elementwise_max_grad XPU kernel to phi,test=kunlun

* move elementwise_min elementwise_min_grad kernels to phi,test=kunlun

* delete elementwise_min_xpu.cc,test=kunlun

* move elementwise_mod elementwise_pow XPU kernels to phi,test=kunlun