* enable infer_ut on windows

* remove lib calculation & time

* unset http_proxy when download bos file on windows

Refactor the organization of layer_norm cuda impl so that it can be reused in fused attention op.

    Extract the layer_norm cuda impl form layer_norm_op.cu to layer_norm_kernel.cu.h.
    Define fused/attention_layer_norm.h, which can be used in fused attention op in next PR.

* Support getitem by Bool index

* delete some debug info of bool index

* support the case that the shape of bool index is different from indexed tensor

This reverts commit 6bacfb0e.

* adamw support cuda

* adamw support cuda

* add rmsprop npu

* add argsort npu

* add argsort npu

* modify according to review

* modify sharedatawith according to review

* modify reshape according to review

* rm dygraph=false

* [NPU] Support npu kernel for pad3d op

* fix for comment of zhouwei25

* fix some bugs according to qili93's comments

* add support and test for paddings in input

* delete VLOG used for debug

* use spin lock in auto growth allocator, test=develop

* use pthread spin lock, test=develop

* use lock guard, test=develop

* use malloc spin lock, test=develop

* use lock_guard, test=develop

* add trainer desc config to distributed strategy

* code style modified

* data_feed set lod

* add depthwise_conv2d npu

* add some tests

* Delete test_unique_op_npu.py

* delete trans input

* [NPU] Support npu op where and where grad

* fix use const_cast

* delete a test

* add (N,C,*) input support for GroupNorm

* --amend

* add npu sin op

* [NPU] Support npu kernel for sin op

* modify support npu kernel for sin op

* modify support npu kernel for sin op

* modify nou sin op

* modify npu sin op

* add sin op npu

* add slim resnet50 quant model in pr-ci-inference

* enable resnet50_quant multi_thread4_trt_int8_bz1

* remove LOG(FATAL)

Add dimension check for inverse to avoid dividing by 0 error when input's shape is [0, 0, 0]. (#34996)

* fix batch_norm and instance norm when input is []

* notest;test=gpu-inference

* notest;test=gpu-inference

* notest;test=gpu-inference

* notest;test=gpu-inference

* fix error

* notest;test=gpu-inference

* notest;test=gpu-inference

* notest;test=gpu-inference

* test=gpu-inference

* add device_context

* add gtest for device_event_gpu

* Remvoe duplicate DeviceType

* push for test

* add unittest

* fix macros

* fix MSVC using usage

* [npu]add rmsprop op

* Add NPU kernel for norm Op: float16 and float32

* fix code for code review

* fix for code review

* add type for paddle_throw

* remove unnecessary head file.\nAdd more testcase

* remove a broadcast

* fix pad outliers err

* fix pad api input type and doc

* fix example of pad

* add unittest for pad3d

* fix unittest

* fix error format

* fix pad doc

* code refactoring, test=develop

* refine, test=develop

* refine, test=develop

* refine, test=develop

* test=develop

* test=develop

* test=develop

* test=develop

[Hybrid Performance] Move the cast op of AMP which cast fp32 param to fp16 param to the optimizer (#34965)

* fix ext_tensor.cast failed bug

* remove useless deps

* fix windows cmake failed

* try to fix windows make failed

* fix make error on windwos

* Add function to disable paddle signal handler

Paddle used google::InstallFaultSignalHandler to handle selected system signals,
mainly for debugging and bug report purposes.

However, this can be conflicted with other python packages whoever captures similar signals.
Such python package involves tvm and more

To resolve this issue, we support a function to disable signal handler

* Remove signal test from WIN32 platform

* Remove redundant return from disable_signal_handler() function

* Add detailed messages to en_doc

* add retry for HcclGetRootInfo

* refine code

* reduce retry interval

* support class center sample of PartialFC