* [ROCM] add rocm 3.5 and 3.9 dockefile, test=develop

* [ROCM] update rocm bin path, test=develop

* [ROCM] add dockerfile for rocm 3.5 amd 3.9, test=document_fix

* [ROCM] fix code stype failure, test=document_fix

* fix fleet for multi-stream

* fix memcpy for ncclid

* use sync to solve move operation

* Add approval for PR-CI-OP-benchmark, test=develop

* dont show token in log, test=document_fix

* merge amp related function in Momentum from paddle.fluid.contrib.optimizer into paddle.optimizer.

* Add unittest for 2.0  Momentum API.

* fix some bugs in weight_decay.

4 APIs: array_length, array_read, array_write, create_array

Add Retry Logic to CublasHandlerHolder to avoid random unittest failure.

* fix flops

* fix flops

[Dy2Stat] Fix bug for loop: a variable is used and created in loop, but used before created (#29769)

* add double grad for conv_transpose

* Test compilation time with less parallel count, notest, test=windows_ci

* optimize rules of Unity Build, notest, test=windows_ci, test=windows_op

* limit parallel counts used only on GPU, test=develop

* remove limit of argument /m:8 on Windows, test=develop

* - Reimplemented elementwise_add grad

- lint

* - fix after review

* - Fix to fix after review

* upgrade oneDNN with GRU INT8 optimizations

* fix test

Enable jit.save to Save Without Running.

Support to transformfor ele in var stms in which var is a slice of Tensor.

* delete no need to calculate inputs in dygraph op_test

* delete no need to calculate inputs in dygraph op_test

Before this commit, test_slice use old api `dygraph_to_static_func` to use Dynamic-t-Static and use Executor explicitly，which is not recommended to users.
After fixed, use recommended API `paddle.jit.to_static` to replace `dygraph_to_static_func`, which won't trigger the random exception on coverage CI.

fix isfinite_v2_op OpProtoAndCheckerMaker AddComment bug 

Unit test reported memory not enough on CPU machines. Reduce batch size again.

* update to_tensor en docs

* add conj op for complex types

* add conj for complex types

* add more test case

* add conj_op test

* modify conj api and impl

* add complex type for fill_constant_op xpu

* add setConstant for complex type

* remove complex conj test file

* user define grad for test_conj_op

* add test case for static mode of conj api

* modify conj doc

* change input args name to x

* remove useless codes

* conj support real types

* add conj test case for real number

* enable_use_gpu has higher priority than FLAGS

* update.

* Windows generate pdb and dump, for debug

* fix code style, test=develop

* modify cmakelist

Modify CublasHandleHolder from using PADDLE_ENFORCE_CUDA_SUCCESS to PADDLE_RETRY_CUDA_SUCCESS to fix random unittest failure. We checked that the unittest log showed CUDA allocation error at this file, which may due to GPU not enough. We fixed similar failure in the past, so we applied PADDLE_RETRY_CUDA_SUCCESS here.

* add complex real op & api & unittest

* add imag op & api & unittest

* refactor op impl

* revert simplify writing due to complile failed

* polish details

* polish grad op code

* update EarlyStopping doc

* update EarlyStopping doc, test=document_fix