* relu forward opt

* add gelu functor

* optimize code

* mean first version

* fix scalar mean

* add fp16 dtype for api

* add pten conj kernel

* modify conj_kernel file path

* add defined cuda macro to cuda/conj_kernel.h

* support FP16 for more ops

* add amp list tests

* refine reduce_mean_grad

* fix OP benchmark ci

* fix fp16 reduce_mean

* updat ut, but still have some problems

* remove mean/reduce_mean fp16 kernel

softmax_with_cross_entropy optimization with soft label. This PR includes optimization of
    "SoftmaxWithCrossEntropySoftLabel" : compute log_softmax and then compute loss.
    "CrossEntropySoftLabel" : compute loss with softmax as input.
These optimization includes following technics:
    read data to buffer with vectorization
    compute max and sum in warp
    fixed loop size with macro
Performance (computation time):
    softmax_with_cross_entropy_0 (forward) : -40.1%
    softmax_with_cross_entropy_0 (backward): -41%

* fix bugs when run reshape

* fix ci bug

* add complex op and `paddle.complex`.

* support multi precision update for LAMB

* hide some api

* fix ci uts

* fix lamb output of dygraph

* remove some changes to some PR

* try to fix Py3 CI compile error

* fix test_imperative_optimizer, add lars ut, add layer_norm ut

* fix ut, fix format

* fix ut

* fix windows ci

* modify sum mean args

* add GetExpectedPtenKernelArgs for redcue_op

* modify kernel args number

* modify kernel args number

From --ptxas-options=-v, SegmentOpsKernel uses 66 registers in a block.
There are two ways to resolve this problem:
    Reduce the threads per block launch configuration
    add __launch_bound__ to give information to nvcc compiler for reducing registers usage
this PR chooses __launch_bound__ solution because changing gpu_launch_config may affect other ops.

* fix_bugs_for_elementwise_branch_selection

* fix merge_dims bugs

* fix all influenced file

* Faster implementation of CPU kernel for ROI_ALIGN Operator

* Add missing variable to CUDA roi_align_op

* Style

* Fix boundaries

* Rename variables for indexes calculation

* Remove unnecessary emplace

* Revert "Remove unnecessary emplace"

This reverts commit c10e87f7fb812f1a672fde32f2690a97d47e2f5a.

* Style

* add activation

* update activation_op

* add unitest for activation

* fix acosh for init, test=develop

Revert "modify the fix_seed attribute in dropout op is a def attribute.test=develop (#38100)" (#38127)

This reverts commit f44add7b.

* add register_ctx_kernel and move scale kernel

* polish details by reviewer comment

* fix xpu compile failed

* fix cmake error

* trainer_device fix and checknan tool for psgpu;test=develop

* disable show_one_table;test=develop

Add elementwise_fmax and elementwise_fmin operators

Add key_padding_mask and attn_mask in sparse_attention Api

1.Key padding mask is a tensor with dimensions [batch_size, seq_len], and attention mask is a tensor with dimensions [seq_len, seq_len]. The data types of the two masks are consistent with Q, K, and V, which are float32 or float64. If the value in Mask is 0, it means that the position needs to be masked.

2.The changed files are mainly paddle/fluid/operators/sparse_attention_op.cu and python/paddle/fluid/tests/unittests/test_sparse_attention_op.py. sparse_attention has three parts: sddmm, softmax, and dsd. Adding the mask operation only needs to modify the softmax. It has no effect on the other two parts. In addition, in order to test the mask function, related tests has been added.

* Add the transformop parameter in TensorReduceFunctorImpl

* Reduce reshape kernel functions in pten

* delete notes

* fix bugs when compile

* modify register name

* fix compile bugs

* Add float16 type for scatter op.

* Add fp16 test for scatter op.

* Add int and int64 support for scatter_grad on gpu.

* Add int and int64 for check_variable_and_dtype routine.

* Minors.

* Code format.

test=document_fix

As the title.

* replace with pten kernel in cast cuda compute and remove unused codes

* rm unused header file

* replace CastCUDAOpKernel with CastOpKernel

* add map_matmul passes to quant2_int8_mkldnn_pass

* fix fc+act fuse (activation scale)

* ci fix, c++17 structured bindings not available

* fix ci static check

* add conv_gelu_mkldnn_fuse_pass

* add post ops

* Reduce reshape kernel functions in pten

* delete notes

* fix bugs when compile