* rename multi_tensor_adam to fused_adam

* fix some bugs

* fix CI coverage

* rename test_fused_adam.py

* fix some bug

* add test_fused_adam_op.py

* fix some bugs

* fix fused_adam_op.cc

* fix CI bugs

* fix CI bug

* fix CI bug

* [XPU] update xccl to 1.0.8

* update xdnn. add uint8 for concat and split.

* update xdnn to 20230215.

* move variable_utils from phi_api_utils to fluid

* fix coment

* update include

* fix bugs

* fix bugs

* fix bugs

* fix bugs

* fix bugs

* update

* update

* fix CI-Windows-OpenBLAS

* fix bugs

* fix bugs

* fix bugs

* update include

* move variable_utils to phi_utils

* fix namespace

* make FusedMultiTransformer supports variable-lengths.

* modify ffn2 when cuda_version >= 11.6 because of #49392.

* code style

* delete remove_padding

* polish namespace

* change static_tensor_operants

* polish namespace

* fix copysign compile error on Windows

* fix more files' macro

* fix py::array_t calling bug

* polish code

* init commit

* fix tensor operator*

* fix compile bug

* bug reproduce

* update commit

* polish codes

* fix compile bug

* test begin

* test begin

* compile finish

* restore origin composite_backward_api

* pass local CI

* fix merge error

* fix merge error

* change py_test from GPU->CPU, test custom op

* polish codes, modify prim unittest

* modify prim unittest

* determine phi_tensor_operants location

* polish codes

* add header file

* solve windows unresolved symbol

* fix some CI error

* add overload defination

* fix CI inference and Windows

* polish codes according to reviewers' opinion

* polish codes according to reviewers' opinion

Co-authored-by: Nzhangxiaoci <zhangxiaoci@baidu.com>

* decouple strided_memcpy

* move strided_memcpy

* move strided_memcpy to phi

* fix namespace

* update

* fix gpu compile bugs

* add multi_tenosr_adam

* update multi_tensor_base.py, test_multi_tensor_adam.py, adamw.py

* fix adam.py optimizer.py

* fix adamw.py

* fix test_multi_tensor_adam.py

* fix CI bug

* fix CI coverage

* fix ci bug

* fix betapow

* fix some bugs

* fix test_adamw_op.py

* fix CI coverage

* fix multi_tensor_adam_kernel.cc

* fix CI bug

* fix multi_tensor_adam_op.cc and test_multi_tensor_adam.py

* fix code style

* update C++ parts

* remove python parts modification temporarily

* add C++ ut

* update betapow copy code logic

* fix ci ut

* fix windows ci

* fix coverage ci

* improve coverage rate

---------
Co-authored-by: Nsneaxiy <sneaxiy@126.com>

* fix bad alloc exp error

* remove needless files

* add dump_walk_path  (#193)

* add dump_walk_path; test=develop

* add dump_walk_path; test=develop

* add dump_walk_path; test=develop

* Add multiple CPU communication, parameter query and merging functions, support batch alignment between multiple cards (#194)

* compatible with edge_type of src2dst and src2etype2dst (#195)

* do not merge_feature_shard when using metapath_split_opt (#198)

* support only load reverse_edge (#199)

* refactor GraphTable (#201)

* fix

* fix

* fix code style

* fix code style

* fix test_dataset

* fix hogwild worker

* fix code style

* fix code style

* fix code style

* fix code style

* fix code style.

* fix code style.

---------
Co-authored-by: Ndanleifeng <52735331+danleifeng@users.noreply.github.com>
Co-authored-by: Nqingshui <qshuihu@gmail.com>
Co-authored-by: NWebbley <liwb5@foxmail.com>
Co-authored-by: Nhuwei02 <53012141+huwei02@users.noreply.github.com>

* check tensor numel in PyObject_CheckLongOrToLong

* add unittest

* fix to_dlpack for loop

* fix reference count

* refactor dir for prim

* support blacklist for bwd comp

* fix type error

* remove additional file

* fix git ignore

* add more test

* merge develop

【PRIM】Support use operator's output metadata info  in constructing static backward composite (#50043)

* [prim] support custom target_gradients

* support infershape after append one gradop

* [prim] add simple net test

* fix test_loop segment fault bug

* [prim] fix infer shape segment fault bug when output of grad_op_desc is empty

* polish static grad op maker gen

* fix some bugs

* fix static code gen

* solve conflict

* modify composite grad maker name

* integrate phi and fluid info in static code gen

* rename some composite maker

* modify static code gen format

* add FP16 dtype for CastNumpy2Scalar

* fix throw message

* add test

* fix SyntaxWarning

* test skip for float16

* fix dtype mistakes

* [CustomDevice] add custom device api

* update

* update

* test=document_fix

* update

* update

* add  examples

* rm flags_retain grad in pybind

* retain grads for xpu test

* set retain grad for xpu

* rm flag

* lint

---------
Co-authored-by: Nwanghuancoder <wanghuan29@baidu.com>

* Support stream priority for standalone executor

* Fix compile error

* Fix compile error

* Fix compile error

* Fix compile error

* Fix compile error

* add set slot_num for psgpuwraper (#177)

* add set slot_num_for_pull_feature for psgpuwarper

* Add get_epoch_finish python interface (#182)

* add get_epoch_finish interface

* add return

* delete return

* add unzip op (#183)

* fix miss key for error dataset (#186)

* fix miss key for error dataset

* fix miss key for error dataset
Co-authored-by: Nyangjunchao <yangjunchao@baidu.com>

* add excluded_train_pair and infer_node_type (#187)

* support return of degree (#188)

* fix task stuck in barrier (#189)
Co-authored-by: Nyangjunchao <yangjunchao@baidu.com>

* check node/feature format when loading (#190)

* check node&feature format when loading

* check node&feature format when loading (2£ (2)

* degrade log (#191)

* [PGLBOX]fix conflict

* [PGLBOX]fix conflict

* [PGLBOX]replace LodTensor with phi::DenseTensor

* [PGLBOX]fix gpu_primitives.h include path

* [PGLBOX]from platform::PADDLE_CUDA_NUM_THREADS to phi::PADDLE_CUDA_NUM_THREADS

* [PGLBOX]fix unzip example code

* [PGLBOX]fix unzip example code

* [PGLBOX]fix unzip example code

* [PGLBOX]fix unzip example code

* [PGLBOX]fix unzip ut

* [PGLBOX]fix unzip ut

* [PGLBOX]fix code style

* [PGLBOX]fix code style

* [PGLBOX]fix code style

* fix code style

* fix code style

* fix unzip ut

* fix unzip ut

* fix unzip ut

* fix unzip

* fix code stype

* add ut

* add c++ ut & fix train_mode_ set

* fix load into memory

* fix c++ ut

* fix c++ ut

* fix c++ ut

* fix c++ ut

* fix code style

* fix collective

* fix unzip_op.cc

* fix barrier

* fix code style

* fix barrier

* fix barrier

* fix code styple

* fix unzip

* add unzip.py

* add unzip.py

* fix unzip.py

---------
Co-authored-by: Nchao9527 <33347532+chao9527@users.noreply.github.com>
Co-authored-by: NSiming Dai <908660116@qq.com>
Co-authored-by: Nhuwei02 <53012141+huwei02@users.noreply.github.com>
Co-authored-by: Nyangjunchao <yangjunchao@baidu.com>

* remove max_slot_num

* fix test case

* refine munmap freq for ref_cnt_mmap_allocator

* add shm reuse logic

* fix compile bug

* fix compile bug

* fix bug of file refcount

* fix compile bug

* fix compile bug

* refine code for delete shm case

* polish code

* refine shm cache pool size setting logic

* set buffer is 2

* refine shm cache size logic

* refine max shm cache

* refine shm cache size

* change feed_op to phi kernel

* fix ci bugs

* fix build bugs

* fix ci bugs

* fix compile bugs

* fix ci bugs

* perfect code

* perfect comment code

* fix install bugs

* modify code according comment

* remove visitor in feed_op

* modify according comment

* perfect code according comment

* add infershape

* fix py3 bugs

* fix getexpected kernel type

* fix getexpected kernel type

* fix ci bugs

* add registry for custom device

* fix py3 bugs

* fix floating point error

* fix py3 test bugs

* Support call backward() without params in dy2st

* clear ProcessGroupCustom manually

* fix bug

* fix bug

* move destroy ProcessGroup to ProcessGroupIdMap

* enable destroy to all device

* remove unused comments

* change to internal api

* Update process_group.cc

* Update process_group.cc

* update get_device to custom

* add custom_device api

* rm is_compiled_with_custom_device from framework

* add todo comments

* Fix reduce func bug in process_group_bkcl

Also catch up with a recent process_group PR that failed to add XPU branch.
Note that reduce is still accomplished by allreduce for xpu. Fix this should
xccl lib be updated.

* fix compile issue for non-XPU