* decouple_memory

* perfect memory utils

* fix ci bugs

* fix inference bugs

* fix custom test bugs

* fix converage bugs

* modify code according comment

* modify namespace

* deal with compile bugs

* Optimize the ernie inference performance on xpu

* fix enable runtime cache logic

* when op's input shape has changed, should create a new runtime context

* fix

* set flag when input shape has changed

[XPU] add multi_encoder_xpu_slice_fuse_pass, generate_sequence_xpu_fuse_pass, generate_sequence_xpu kernel (#50570)

* Add matmul_v2 to the quantization process and adjust Ernie model test

* Correct cpu_quantize_pass test

* Move op to fuse transformation to placement pass

* Correct test

* support xpu multi-card infer

* add ut

* clean code

* clean code

* fix

* fix

* fix

* fix

* expand mix_precision to custom_device

* fix bug

* fix bug

* fix comment

* fix DEFINE bug

* add cast setvalue op

* add set_value to op teller

* renew test and add description

* add setAxis and add complex test

* change test

* eleadd_trans first version

log fix

* refine code for linear format, add pass check

* linear format refine and ut fix

* fix ut

* windows ut

* windows ut 2

* move tensorMeta and alloc to configure

* update

* support int64 shape tensor as engine input

* add inference_predictor ut

* Adjust mkldnn_placement_pass to check library type and data type

* Check if var has inputs

* Remove unrelated test

* Refactor

* trans_layernorm

* QuantTranpose pattern is being found by pass

* quant + transpose fuse

* code style changes

* UT written, reorder fixed

* Dequantize + transpose2 fuse  added

* pass name changed

* UT added & shift corrected

* got rid of redundancy

* review changes

* AsIntermediate corrected

* compat added

* disable conv2d_fusion_layout_transfer_pass temporarily

* disable conv2d_fusion_layout_transfer_pass temporarily

* preln_residual 2 fused_bias_residual

* skip layernorm fix and ut

* code refine

* code style refine

* fix ut

* fix output

* add trt layer fall back info

* refine op teller and ut

* DropoutMaskOut output fix

* jit layer support multi thread

* fix bug

* clone prediector not do graph optimizer

* format

* fix comment and format

* fix override and fromat

* fix

* fix

* gn_silu

* add ut

* set TIMEOUT

* correct comments

* comments

* disable windows ut

* rename parameter

* support PaddlePaddle Backend on Triton

* fix test cases

* fix Codestyle

* add test case

* add test case

* change feed_op to phi kernel

* fix ci bugs

* fix build bugs

* fix ci bugs

* fix compile bugs

* fix ci bugs

* perfect code

* perfect comment code

* fix install bugs

* modify code according comment

* remove visitor in feed_op

* modify according comment

* perfect code according comment

* add infershape

* fix py3 bugs

* fix getexpected kernel type

* fix getexpected kernel type

* fix ci bugs

* add registry for custom device

* fix py3 bugs

* fix floating point error

* fix py3 test bugs

* add trt_support_nhwc_pass

* add gpu_cpu_map_matmul_to_mul_pass to kGpuLowerPrecisionPasses

* disable fc_elementwise_layernorm_fuse_pass in mixed precision

* add fmha_flashattention oss plugin

* fix paddle_infer_contrib include

* skip_groupnorm

* init

* preln

* add ut

* more assert

* set timeout

* fix windows ci issue

* Implement conv2d_fusion NHWC format using CUTLASS
* Add unit testing for CUTLASS Conv in inference
* Add experimental API for CUTLASS.

* refactor conv_activation_mkldnn_fuse_pass

* refactor conv_affine_channel_mkldnn_fuse_pass

* fix conv_activation_mkldnn_fuse_pass

* fix mkldnn unittest

* refactor int8_scale_calculation_mkldnn_pass and params_quantization_mkldnn_pass

* refactor conv_elementwise_add_mkldnn_fuse_pass

* fix quant

* refactor conv_bn_fuse_pass

* fix conv_bn_fuse_pass

* refactor depthwise_conv_bn_fuse_pass

* fix unittest

* fix conv_bn_fuse_pass

* remove redundant conv2d in params_quantization_mkldnn_pass

* fix params_quantization_mkldnn_pass_tester