add index initialization in the block loop for index_sample kernel when dealing with a input tensor whose shape is larger than block_dim * grid_dim (#39736)

* add block and grid loop for index_sample kernel to deal with a large-shape tensor

* fix code format

* limit grid dim

* fix the omissive initialization of index_i in the second cycle for index_sample kernel

* fix conflicts

* add more op int16 support

* fix xpu ci

* Unify paddle/pten::framework::ddim into pten::ddim

* fix paddle namespace

* compile sucessfully

* fix npu src file

* fix conflict

* fix conflict

* fix tensorrt compiler error

* fix conflict

* fix conflict

* fix tesst file conflict

* fix conflict

* fix mlu file conflict

* fix mlu file conflict

* fix cinn header file conflict

* fix conflict

* fix conflict

* fix conflict

* fix conflict

* Add selected_rows kernel for full

* remove fill_constant register in fluid

* fix bug without GPU

* add jit_kernel_helper dependency for fc

* do some refactor

* add unittest for ops signatures

* add coverage unittest

* fix merge conflict

* fix full selectew_rows bug

* fix RecordEvent interface

* modify default level to 4

* update interface use

* add const default trace level

* update record event interface using

* update operator.cc

* update part1

* fix include profiler.h header in ps server

* fix include profiler.h header in ps server

* Enabled test_matmul_v2_op for final state Eager Dygraph

* Fixed minor issue

* Fixed format issue

* support parse cc in gpu

* change file name

* fix RecordEvent interface

* modify default level to 4

* update interface use

* add const default trace level

* update operator.cc

* remove manual_api

* change sig map of full and empty

* fix fill_any_like_xpu_op

* fix fill_any_like_xpu_op

* fix problem of fill_any_like_xpu_op

* fix conflict

* polish code

* [Eager] Support GradientHook before running seperate GradNode

* Fix CI issue

* Support eager ReduceHook in accumulation_node

* Fix CI  issue

* Add some tests to fix coverage CI issue

* add DistributedFusedLamb op

* polish code

* fix compile error

* compatible with pten changement

* fix rocm compile error

* improve converage

* update upstream/develop

* fix cast_with_ptr.h

* add FLAGS_distributed_lamb_divide_nranks_when_allreduce=1

* fix clip before allreduce

* add use_master_param_norm

* code polish

* fix bug

* fix ROCM ci

* merge legacy to fluid

* Remove legacy code

* Remove legacy code

* Remove DataType test

* Using Tensor directly instead of using EagerTensor

* support gradient_accumulation

* make test_imperative_lod_tensor_to_selected_rows longer

* make test_imperative_lod_tensor_to_selected_rows longer

* refine code

* Rename all EagerTensor to Tensor

* Rename some EagerTensor to Tensor

* rename EagerTensor to EagerVariable

* add more test

* Support copiable selected rows and merge develop

* move blas related files
* move lapack related files

* add cinn_instruction_run_op test code

* update several interfaces of CinnLaunchContext

* update several interfaces and add detail comments in CinnLaunchContext class

* to skip the bug of error message check

* fix ut test failed due to reliant interface updated

* trans diagonal kernel into pten

* fix by code review

* support dtype param for auto_cast

* add amp_dtype for tracer

* add unsupported bf16 list

* support bf16 amp for O2

* refine python interface for bfloat16

* refine code

* refine code

* refine unittest

* refine code

* refine code

* add bf16 o1

* refine code by comment

* add gradient accumulator

* add recompute

* Update IpuStrategy

* fix ci

* rerun ci

* fix_sharding_group

* fix_sharding_group

* new way of unit test , *test=kunlun

* new way of ut, *test=kunlun

* the mlir representation of pten, test=develop

* fixes an error, test=develop

* infrt registers pten kernels
Co-authored-by: NShixiaowei02 <39303645+Shixiaowei02@users.noreply.github.com>

* support inplace and intermediate in yaml

* add cmake for dygraph_api

* add T to Alloc and remove default value of DataType in DeviceContext::Alloc

* add dtype

* add tool: print kernel signaturs

* fix windows compile

* [MLU]add matmul and matmul_v2 op

* [MLU] fix data_type and del matmul

* [MLU] fix compile error

* [MLU] fix ci_check error

* [MLU]add sync stream ops and broadcast pytest

* [MLU]fix broadcast pytest to add data type

* merge legacy to fluid

* Remove legacy code

* Remove legacy code

* Remove DataType test

* Using Tensor directly instead of using EagerTensor

* support gradient_accumulation

* make test_imperative_lod_tensor_to_selected_rows longer

* make test_imperative_lod_tensor_to_selected_rows longer

* refine code

* Rename all EagerTensor to Tensor

* Rename some EagerTensor to Tensor

* rename EagerTensor to EagerVariable

* add more test

* fix different device gradient_accmulator bug

* merge develop

* remove useless tests

* [Eager] Support GradientHook before running seperate GradNode

* Fix CI issue

* Fix CI issue

* pten matmul cuda kernel support bf16

* avoid custom kernel deps on pten_function_api

* Revert "pten matmul cuda kernel support bf16"

This reverts commit 5d520845b9a189375677276efb673235ed8e5ee0.

* refine code

* fix compile

* fix test_split_api

* move bernoulli kernel to pten

* follow comments

* relocate code of interpretercore gc