* optimize content-dnn cuda kernel

* Support Mask RCNN2 (#2588)

* add reshape opencl kernel && optimise conv 1x1 ,test=develop

* add reshape opencl kernel && optimise conv 1x1 &&code style ,test=develop

* add reshape opencl kernel && optimise conv 1x1 &&code style ,test=develop

* fix: update backend fpga patch

[LITE][OPENCL] Add relu image2d kernel unit test, Fix conv2d_1x1, relu, layout using new Image2D Layout (#2564)

* add 3 layout for opencl image. test=develop

* add relu image2d test. test=develop

* [LITE][OPENCL] Add depthwise_conv_3x3 opencl kernel

* [LITE][OPENCL] Add depthwise_conv_3x3 opencl kernel. test=develop

* [LITE][OPENCL] Add Pool opencl kernel. test=develop

optimize search_grnn

[LITE][NPU][XPU] Refine subgraph pass, and support NPU/XPU model generation at execution time (#2576)

* add opencl conv1x1 image impl and unit test pass with relu & bias,
add layout_compute --> buffer2image float32 --> with unit test pass
suite checked test for more situation , test=develop

* add opencl conv1x1 image impl and unit test pass with relu & bias,
add layout_compute --> buffer2image float32 --> with unit test pass
suite checked test for more situation , test=develop

* fix white space cpp lint , test=develop

test=develop

* add arm split lod tensor, test=develop

* add arm merge lod tensor, test=develop

* update split merge lod tensor, test=develop

* add reduce_prob op, test=develop

* support mask_rcnn succeed, test=develop

* init resnet cuda int8 support
test=develop

* refine cuda unit test
test=develop

* add the forgeted file.
test=develop

* feat: add sgemm4x4 for armv8

* fix: fix armv7 gemm choose condition

test=develop

fix: fix conv_block prepack_input_nxwc4 bug
* fix: optimize sgemm_c4 in armv7
     change condition of choose winograd kernel
* fix: change conv choose kernel condition

* unfinish sgemmc4

* finish armv8 sgemmc4

* arm add sgemmc4 with deal with remain

* [ARM] add sgemmc4 small kernel, test=develop

* [ARM] sgemmc4 small improve armv7 speed by add 4x8 block, test=develop

* fix conv 2-pad to 4-pad

* fix compute conv shape

* fix pad, test=develop

* change conv_depthwise_3x3s1_fp.cc name to conv3x3s1p01_depthwise_fp32.cc to distinguish between conv3x3s1_depthwise_fp32.cc

* delete printf note in conv3x3s1, test=develop

* delete printf note, test=develop

* delete gem_sdot.h, test=develop

it is coped from __gemm_sdot_meta_.h

* update compute padding, test=develop

* fix padding size, must be 2 or 4. test=develop

* fix format in operators/conv_op.cc, test=develop

* change #if 0 to #if 1, test=develop

* put 2-pad to 4-pad in AttachImpl, test=develop

* fix clang-format error inn tests/math/connv_compute_test, test=develop

* fix x86 test result error, test=develop

* add asymmetric padding test case in liite/tests/math/conv_compute.cc, test=develop

* change paddings type to support dynamically modify, test=develop

* fix x86 build error in connv_compute_test, test=develop

* fix opencl build error, test=develop

* fix oopencl build error, test=develop

* fix  opencl/conv_compute build error, test=develop

* fix  opencl/conv_compute build error, test=develop

* fix format in kernels/opencl/conv_computte_ttest,test=develop

* fix build error, test=develop

fix build error in kernels/x86/conv_compute.h

* add nhwc to nchw

* add layout in funcs

* change layout as extra, test=develop

* change make, test=develop

* use template class method to update layout NNCHHW and NHWC transform, test=develop

* fix cmake error, set layout to extra, test=develop

* fix test_layout_compute_arm test, its extra

* layout is extra, test=develop

* fix error in kernels/arm/layout_comput.cc when register kernel, DataLayout must be NCHW, test=develop

* delete extra note, test=develop

* delete extra test

* delete layout_test, test=develop

, its in tests/math/layout_comput_test

* delete extrat test, test=develop

* unfinish sgemmc4

* finish armv8 sgemmc4

* arm add sgemmc4 with deal with remain

* [ARM] add sgemmc4 small kernel, test=develop

* fix pooling bug and speed

* fix build error

* delete VLOGin pool, test=develop

* add openmp, test=develop

* fix lite/kernels/arm/pool_compute_test basic_pooling compute error bug, test=develop

* update pooling 2-pad to 4-pad, test=develop

* fix 2-pad to 4-pad in operators/pool_op.h, AttachKernel will set param, so 2-pad to 4-pad funcs should put in AttachKernel. test=ddevellop

* put 2-pad to 4-pad in AttachImpl, test=develop

* according to reviews, fix some format error. test=develop

* fix format errorr, add (). test=develop

* change paddings type to support dynamically modify, test=develop

* update padding type int other devices, test=develop

* fix x8d build error on shared_ptr, test=ddevelop

* fix formmat in operators pool_op.cc, test=develop

* fix cuda building, test=develop

* remove sequence_pool from cmake because build error, test=develop

* [ARM] sgemv support transA, test=develop

* add sgemv ut, test=develop

* add x86 op and kernel : search_fc and sequence_topk_avg_pooling   for content-dnn model test=develop

* Fix bug target for kHost and kARM not equal. test=develop

* Fix license. test=develop

* add debug -g option. test=develop

* enable opencl demo. test=develop

* Fix model_optimize_tool found no opencl kernel. test=develop

* add more vlog. test=develop

* remove macro LITE_WITH_OPENCL, LITE_WITH_FPGA in passes. test=develop

* Fix valid_places in mobilenetv1_test. test=develop

* Fix bug of find no real output of fetch, after tool OPs of optimzer passes. test=develop

* Fix vlog as log message in model_optimize_tool. test=develop

* fix miscs. test=develop

* fix comment. test=develop

* Fix misspell of opencl, fpga kernels name in lite/api/CMakeLists.txt. test=develop

* add opencl macro in full_api of demo. test=develop

* add cuda match_matrix_tensor op and test, test=develop

align the lite nearest， bilinear op to fluid on arm and cuda

* fix pooling bug and speed

* fix build error

* delete VLOG in pool, test=develop

* add openmp, test=develop

* fix lite/kernels/arm/pool_compute_test basic_pooling compute error bug, test=develop

[LITE][NPU] Add fusion_elementwise_add_activation bridge, remove the limitation of the dimensions of input tensors in the graph compute kernel, and refine log message (#2395)

test=develop

fix jit::matmul bug. Input x shape is (m, k), weight shape is (k, n). When k < 512, m==1, and n is a multiple of 16, the weight pointer is not correctly updated in the group calculation in the implementation of jit::matmul, resulting in the result diff