support pure bfloat16 for more ops

* Support both use_calc_stream and sync_op in send recv APIs (#46023)

* Support both use_calc_stream and sync_op in allgather API (#46295)

* Support both use_calc_stream and sync_op in collective communication API (#46761)

* Move group and all reduce from collective to communication (#45848)

* Completes bfloat16 dtype for collective api in eager mode (#45844)

* Fix collective APIs cannot be recognized when building docs (#46962)
Co-authored-by: NLiYuRio <63526175+LiYuRio@users.noreply.github.com>

* paddle-inference support custom-ops
Co-authored-by: NZhixin Yao <zhixiny@graphcore.ai>

* fix tolower
Co-authored-by: NZhixin Yao <zhixiny@graphcore.ai>
Co-authored-by: NZhixin Yao <zhixiny@graphcore.ai>

* [PHI] Migrate gelu kernels (#45596)

* gaussian random

* mkldnn to onednn renaming

* fix merge conflicts

* remove fluid code

* onednn renaming

* gelu fwd

* sort activations

* gelu gradient

* remove unused macros

* merge conflicts

* fix merge conflicts

* remove extra contraint from gelu op

* [PHI] relu6_grad kernel (#46501)

* Relu6

* remove fluid handler

* add individual kernel signature

* coding style

* replace bounded_relu with clip

* whitespace

* code style

Co-authored-by: NRichardWooSJTU <37864677+RichardWooSJTU@users.noreply.github.com>

* cherry pick delay tensorrt log
* Update trt_plugin.h

* [CustomDevice] add dy2static support

* update

* add gemm_epilogue

* xpu-paddlepaddle-40 [任务] fused_gemm_epilogue 支持 test=kunlun

* update protobuf format

* fix protobuf content

* fix file mode

* fix compiling error when gpu not exists

* fix compiling error when gpu not exists

* fix compiling error when gpu not exists

* fix compiling error when gpu not exists

* support rocm

* gaussian random

* mkldnn to onednn renaming

* fix merge conflicts

* remove fluid code

* onednn renaming

* Move classes from mkldnn_reuse.h to onednn_reuse.h

* Move more functions from mkldnn_helper.h to onednn_helpper.h

* Change MKLDNN to OneDNN in VLOG message
Co-authored-by: NSilv3S <slawomir.siwek@intel.com>

test=kunlun

* refine cmake of framework

* add deps for dense tensor

* fix deps

* remove alloc(ctx)

* add depends on mkldnn

* abs relu6 fwd

* abs bwd

* gaussian_random_kernel and mkldnn-onednn renaming

* scale kernel

* whitespace

* whitespace

* revert scale migration

* whitespaces

* revert changes to gaussian kernel

* whitespaces

* support depthwise_conv2d ops
Co-authored-by: NZhixin Yao <zhixiny@graphcore.ai>
Co-authored-by: NZhaorui Chen <zhaoruic@graphcore.ai>

* fix duplicate name
Co-authored-by: NZhixin Yao <zhixiny@graphcore.ai>
Co-authored-by: NZhaorui Chen <zhaoruic@graphcore.ai>

* optimizer conv alog speed

* code polish

* remove useless code

* fix compile error

* fix cpu compile error

* not use cudnn alog t

* add search cache max number

* polish code

* fix cache test bug

* add groups data format to conv args

* fix cache test bug

* fix cudnn_deterministic bug

* fix test switch auto tune bug

* fix test swith autotune bug;

* fix conv cache bug

* fix cache test error

* fix cache test bug

* fix windows mac compile error

* fix workspace search error

* update cudnn cache

* fix cache test bug; test=develop

* fix autotune swith test error

* polish code

* oplish code

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support fp16 of adam operator in xpu environment. test=kunlun

* support fp16 of adam operator in xpu environment. test=kunlun

* support fp16 of adam operator in xpu environment. test=kunlun

* [CustomDevice] add profiler apis

* migrate CalculateEstOccupancy into cuda_tracer

* update

* add ut

* Add int8 support for matmul+elementwiae_add fuse

* Corrections after review and ernie test fix

* add merged_momentum *test=kunlun

* add merged_momentum *test=kunlun

* add fp16 to merged_momentum,*test=kunlun

* change dist_model.cc

* add merged_momentum unittest and  change momentum,test=kunlun

* add merged_momentum unittest and  change momentum,test=kunlun

* add merged_momentum unittest and  change momentum,test=kunlun

* add merged_momentum unittest and  change momentum,test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* support beam_search operator on xpu. test=kunlun

* fix beam_search operator bugs on xpu. test=kunlun

* fix beam_search operator bugs on xpu. test=kunlun

* fix beam_search operator bugs on xpu. test=kunlun

* fix beam_search operator bugs on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* support beam_search_decode operator on xpu. test=kunlun

* change to async mode for xpu multi-card training in static graph mode

* minor bugfix

* irrelevant. move to another pr

* move change to other pr

* fix stream issue

* fix 'stream not meet with current context' error

* fix branch diverge, test=kunlun

* xpu unittest grad compute supports more types, *test=kunlun

* add instance norm xpu, *test=kunlun

* [XPU] add some collective ops. test=kunlun

* use XPUOpTestWrapper. test=kunlun

* skip kl1 for collective ops. fix typo: deivce -> device. test=kunlun