提交 6129ab42 编写于 作者: Y Yi Wang 提交者: GitHub

Merge pull request #2964 from Canpio/dev_refactor_tensor

Simplify Tensor implementation
上级 ef28f66f 06358129
develop 2.0.1-rocm-post Ligoml-patch-1 OliverLPH-patch-1 OliverLPH-patch-2 PaddlePM-patch-1 PaddlePM-patch-2 ZHUI-patch-1 add_default_att add_model_benchmark_ci add_some_yaml_config addfile all_new_design_exec ascendrc ascendrelease cherry_undefined_var compile_windows delete_2.0.1-rocm-post delete_add_default_att delete_all_new_design_exec delete_ascendrc delete_compile_windows delete_delete_addfile delete_disable_iterable_dataset_unittest delete_fix_dataloader_memory_leak delete_fix_imperative_dygraph_error delete_fix_retry_ci delete_fix_undefined_var delete_improve_sccache delete_incubate/lite delete_paddle_tiny_install delete_paralleltest delete_prv-disable-more-cache delete_revert-31068-fix_conv3d_windows delete_revert-31562-mean delete_revert-33630-bug-fix delete_revert-34159-add_npu_bce_logical_dev delete_revert-34910-spinlocks_for_allocator delete_revert-35069-revert-34910-spinlocks_for_allocator delete_revert-36057-dev/read_flags_in_ut dingjiaweiww-patch-1 disable_iterable_dataset_unittest dy2static enable_eager_model_test final_state_gen_python_c final_state_intermediate fix-numpy-issue fix_concat_slice fix_dataloader_memory_leak fix_imperative_dygraph_error fix_npu_ci fix_op_flops fix_retry_ci fix_rnn_docs fix_tensor_type fix_undefined_var fixiscan fixiscan1 fixiscan2 fixiscan3 github/fork/123malin/netifaces github/fork/123malin/tdm_abacus github/fork/AshburnLee/dev_unique github/fork/ForFishes/fix_memory_matmul github/fork/ForFishes/rm_fluid github/fork/LielinJiang/move-2.0-api github/fork/LielinJiang/visual-dl-cb github/fork/LiuChiachi/add-transformer-generate-square-subsequent-mask-api github/fork/LiuChiachi/fix-example-code-for-hapi-Model github/fork/LiuChiachi/remove-input-requirment-in-dygraph-Model github/fork/MrChengmo/fix_ps_profiler github/fork/MrChengmo/update_ps_heter github/fork/PWhiddy/patch-1 github/fork/Shixiaowei02/dev/save_load_upgrade github/fork/TCChenlong/fix_hapi github/fork/TCChenlong/fix_inden github/fork/Thunderbrook/xpu_slice github/fork/XieYunshen/disable_ut_test_parallel_executor_fetch_isolated_var github/fork/XieYunshen/disable_ut_test_parallel_executor_fetch_isolated_var_2 github/fork/XieYunshen/disable_ut_test_parallel_executor_fetch_isolated_var_3 github/fork/XieYunshen/timeout_20S_ut github/fork/ZeyuChen/remove-nltk github/fork/arlesniak/arlesniak/selective__mkldnn_flags github/fork/baiyfbupt/code_doc_mig github/fork/chalsliu/set_timeout github/fork/chen-zhiyu/develop github/fork/chenwhql/ci/try_to_find_test_buffer_shared_memory_reuse_pass_error github/fork/chenwhql/dygraph/remove_scale_loss_and_apply_collective_grads github/fork/chenwhql/saveload/add_get_inference_program github/fork/chenwhql/saveload/remove_save_load_config github/fork/cryoco/pass-compatibility-trt github/fork/danleifeng/isempty_api2.0 github/fork/frankwhzhang/api_transfer github/fork/hbwx24/error_msg/cuda_kernel_error_msg github/fork/heavengate/cherry_yolo_box github/fork/heavengate/update_yolo_box github/fork/iclementine/rnn_fix github/fork/iducn/testestse github/fork/jczaja/prv-25537-fix github/fork/jeff41404/release/1.8 github/fork/jiweibo/api_2.0 github/fork/jiweibo/fix_lite_resnet50_test github/fork/juncaipeng/fix_doc_1 github/fork/lfchener/sample_code github/fork/littletomatodonkey/fix_reg_doc github/fork/liym27/dy2stat_update_assign_to_rc20 github/fork/luotao1/profiler_ut github/fork/mapingshuo/add_wait github/fork/mapingshuo/doc_2.0 github/fork/mapingshuo/zero-0.5 github/fork/miraiwk/dev github/fork/pangyoki/add-Categorical-class-branch github/fork/pangyoki/add-multinomial-op-branch github/fork/pangyoki/fix-test_distritbution-CI github/fork/qjing666/doublegrad github/fork/qjing666/fix_hdfs_download github/fork/sandyhouse/add_gather_etc github/fork/sandyhouse/add_send_recv_alltoall_etc github/fork/sandyhouse/pipeline_exe_run github/fork/seiriosPlus/feature/large_scale_kv_save_delta github/fork/seiriosPlus/fix/paddle_errors_fix github/fork/seiriosPlus/fix/paddle_op_errors github/fork/shangzhizhou/fix_test_activation_op_random_bug github/fork/smallv0221/yxp0924 github/fork/smallv0221/yxp0925 github/fork/swtkiwi/del-matplotlib github/fork/tianshuo78520a/kunlun_test github/fork/tianshuo78520a/update_dockerfile github/fork/wanghaoshuang/bert_fuse github/fork/wanghaoshuang/label_smooth github/fork/wanghuancoder/develop_CUDASynchronize github/fork/wanghuancoder/develop_Layer_doc github/fork/wanghuancoder/develop_ParameterList_doc github/fork/wanghuancoder/develop_Sequential_doc github/fork/wanghuancoder/develop_bilinear_tensor_product github/fork/wanghuancoder/develop_coverage_build_sh github/fork/wanghuancoder/develop_in_dynamic_mode_doc github/fork/wanghuancoder/develop_unique_name_doc github/fork/wangxicoding/fleet_meta_combine github/fork/wawltor/error_message_fix_5 github/fork/willthefrog/remove_l2_norm github/fork/windstamp/momentum_op github/fork/windstamp/mv_op_5 github/fork/windstamp/normal_api github/fork/wojtuss/wojtuss/fusion_gru_quantization github/fork/wojtuss/wojtuss/quantization-with-shift github/fork/wzzju/fix_err_info github/fork/wzzju/pure_fp16 github/fork/xiemoyuan/op_error_message github/fork/xiemoyuan/optimize_error_message github/fork/yaoxuefeng6/fix_doc github/fork/yaoxuefeng6/mod_dataset_v2 github/fork/yongqiangma/lod github/fork/ysh329/fix-clip-by-norm-error github/fork/ysh329/fix-error-clip-by-value github/fork/yukavio/error_info github/fork/zhangting2020/conv_filter_grad github/fork/zhangting2020/is_compile_with_cuda github/fork/zhangting2020/place_doc github/fork/zhangting2020/program github/fork/zhhsplendid/fix_any github/fork/zhhsplendid/refine_api2 github/fork/zhhsplendid/refine_api2_test github/fork/zhhsplendid/refine_api_test_ptb_lm github/fork/zhhsplendid/refine_api_test_resnet github/fork/zhhsplendid/refine_api_test_simnet github/fork/zhiqiu/dev/refine_initializer github/fork/zhiqiu/dev/remove_inplace_argument github/fork/zlsh80826/nvinfer_plugin_var_len_cuda11 improve_sccache incubate/infrt incubate/lite inplace_addto make_flag_adding_easier master move_embedding_to_phi move_histogram_to_pten move_sgd_to_phi move_slice_to_pten move_temporal_shift_to_phi move_yolo_box_to_phi npu_fix_alloc numel paddle_tiny_install paralleltest preln_ernie prv-disable-more-cache prv-md-even-more prv-onednn-2.5 pten_tensor_refactor release/0.11.0 release/0.12.0 release/0.13.0 release/0.14.0 release/0.15.0 release/1.0.0 release/1.1 release/1.2 release/1.3 release/1.4 release/1.5 release/1.6 release/1.7 release/1.8 release/2.0 release/2.0-alpha release/2.0-beta release/2.0-rc release/2.0-rc1 release/2.1 release/2.2 release/2.3 release/2.3-fc-ernie-fix release/2.4 release/lite-0.1 revert-24981-add_device_attr_for_regulization revert-26856-strategy_example2 revert-27520-disable_pr revert-31068-fix_conv3d_windows revert-31562-mean revert-32290-develop-hardlabel revert-33037-forci revert-33475-fix_cifar_label_dimension revert-33630-bug-fix revert-34159-add_npu_bce_logical_dev revert-34406-add_copy_from_tensor revert-34910-spinlocks_for_allocator revert-35069-revert-34910-spinlocks_for_allocator revert-36057-dev/read_flags_in_ut revert-36201-refine_fast_threaded_ssa_graph_executor revert-36985-add_license revert-37318-refactor_dygraph_to_eager revert-37926-eager_coreops_500 revert-37956-revert-37727-pylayer_support_tuple revert-38100-mingdong revert-38301-allocation_rearrange_pr revert-38703-numpy_bf16_package_reupload revert-38732-remove_useless_header_in_elementwise_mul_grad revert-38959-Reduce_Grad revert-39143-adjust_empty revert-39227-move_trace_op_to_pten revert-39268-dev/remove_concat_fluid_kernel revert-40170-support_partial_grad revert-41056-revert-40727-move_some_activaion_to_phi revert-41065-revert-40993-mv_ele_floordiv_pow revert-41068-revert-40790-phi_new revert-41944-smaller_inference_api_test revert-42149-do-not-reset-default-stream-for-stream-safe-cuda-allocator revert-43155-fix_ut_tempfile revert-43882-revert-41944-smaller_inference_api_test revert-45808-phi/simplify_size_op revert-46827-deform_comment rocm_dev_0217 support_weight_transpose test_benchmark_ci test_feature_precision_test_c test_model_benchmark test_model_benchmark_ci zhiqiu-patch-1 v2.4.0-rc0 v2.3.2 v2.3.1 v2.3.0 v2.3.0-rc0 v2.2.2 v2.2.1 v2.2.0 v2.2.0-rc0 v2.2.0-bak0 v2.1.3 v2.1.2 v2.1.1 v2.1.0 v2.1.0-rc0 v2.0.2 v2.0.1 v2.0.0 v2.0.0-rc1 v2.0.0-rc0 v2.0.0-beta0 v2.0.0-alpha0 v1.8.5 v1.8.4 v1.8.3 v1.8.2 v1.8.1 v1.8.0 v1.7.2 v1.7.1 v1.7.0 v1.6.3 v1.6.2 v1.6.1 v1.6.0 v1.6.0-rc0 v1.5.2 v1.5.1 v1.5.0 v1.4.1 v1.4.0 v1.3.2 v1.3.1 v1.3.0 v1.2.1 v1.2.0 v1.1.0 v1.0.2 v1.0.1 v1.0.0 v1.0.0-rc0 v0.15.0 v0.15.0-rc0 v0.14.0 v0.13.0 v0.12.0 v0.11.1a2 v0.11.1a1 v0.11.0 lite-v0.1
无相关合并请求
......@@ -48,25 +48,27 @@ class Tensor {
template <typename T>
const T* data() const {
CheckDims<T>();
EnforceSufficientMemory<T>();
return reinterpret_cast<const T*>(
reinterpret_cast<uintptr_t>(holder_->ptr()) + offset_);
}
template <typename T>
T* data() {
CheckDims<T>();
EnforceSufficientMemory<T>();
return reinterpret_cast<T*>(reinterpret_cast<uintptr_t>(holder_->ptr()) +
offset_);
}
template <typename T>
template <typename T, // must be POD types
typename std::enable_if<std::is_pod<T>::value>::type* = nullptr>
T* mutable_data(DDim dims, platform::Place place) {
set_dims(dims);
Resize(dims);
return mutable_data<T>(place);
}
template <typename T>
template <typename T, // must be POD types
typename std::enable_if<std::is_pod<T>::value>::type* = nullptr>
T* mutable_data(platform::Place place) {
PADDLE_ENFORCE(product(dims_) > 0,
"Tensor's numel must be larger than zero to call "
......@@ -95,11 +97,9 @@ class Tensor {
}
template <typename T>
void ShareDataFrom(const Tensor& src) {
src.CheckDims<T>();
holder_ = src.holder_;
set_dims(src.dims());
offset_ = src.offset_;
void ShareDataWith(const Tensor& src) {
src.EnforceSufficientMemory<T>();
*this = src;
}
template <typename T>
......@@ -107,9 +107,9 @@ class Tensor {
PADDLE_ENFORCE(platform::is_cpu_place(src.holder_->place()) &&
platform::is_cpu_place(dst_place),
"Tensor::CopyFrom only support CPU now.");
src.CheckDims<T>();
src.EnforceSufficientMemory<T>();
size_t size = product(src.dims_) * sizeof(T);
set_dims(src.dims());
Resize(src.dims());
const void* src_ptr = static_cast<const void*>(src.data<T>());
void* dst_ptr = static_cast<void*>(mutable_data<T>(dst_place));
memcpy(dst_ptr, src_ptr, size);
......@@ -117,34 +117,25 @@ class Tensor {
template <typename T>
Tensor Slice(const int& begin_idx, const int& end_idx) const {
CheckDims<T>();
PADDLE_ENFORCE(begin_idx >= 0 && end_idx <= dims_[0],
"Slice index is less than zero or out of bound.");
EnforceSufficientMemory<T>();
PADDLE_ENFORCE(begin_idx >= 0, "Slice begin index is less than zero.");
PADDLE_ENFORCE(end_idx <= dims_[0], "Slice end index is out of bound.");
PADDLE_ENFORCE(begin_idx < end_idx,
"Begin index must be less than end index.");
PADDLE_ENFORCE(dims_[0] != 1, "Can not slice a tensor with dims_[0] = 1.");
std::vector<int> d = vectorize(dims_);
int base = 1;
for (size_t i = 1; i < d.size(); ++i) {
base *= d[i];
}
int base = product(dims_) / dims_[0];
Tensor dst;
dst.holder_ = holder_;
DDim dst_dims = dims_;
dst_dims[0] = end_idx - begin_idx;
dst.set_dims(dst_dims);
dst.Resize(dst_dims);
dst.offset_ = offset_ + begin_idx * base * sizeof(T);
return dst;
}
void set_dims(const DDim& dims) {
if (dims == dims_) {
return;
}
dims_ = dims;
}
void Resize(const DDim& dims) { dims_ = dims; }
DDim dims() const { return dims_; }
const DDim& dims() const { return dims_; }
private:
// Placeholder hides type T, so it doesn't appear as a template
......@@ -159,21 +150,9 @@ class Tensor {
template <typename T, typename PlaceType>
struct PlaceholderImpl : public Placeholder {
private:
template <typename PType>
class Deleter {
public:
Deleter(PType place) : place_(place) {}
void operator()(T* ptr) { memory::Free(place_, static_cast<void*>(ptr)); }
private:
PType place_;
};
public:
PlaceholderImpl(PlaceType place, size_t size)
: ptr_(static_cast<T*>(memory::Alloc(place, size)),
Deleter<PlaceType>(place)),
memory::PODDeleter<T, PlaceType>(place)),
place_(place),
size_(size) {}
......@@ -182,13 +161,13 @@ class Tensor {
virtual paddle::platform::Place place() const { return place_; }
virtual std::type_index type() const { return std::type_index(typeid(T)); }
std::unique_ptr<T, Deleter<PlaceType>> ptr_;
std::unique_ptr<T, memory::PODDeleter<T, PlaceType>> ptr_;
platform::Place place_; // record the place of ptr_.
size_t size_; // size of the memory block.
};
template <typename T>
inline void CheckDims() const {
inline void EnforceSufficientMemory() const {
PADDLE_ENFORCE(holder_ != nullptr,
"Tenosr holds no memory. Call Tensor::mutable_data first.");
PADDLE_ENFORCE(holder_->size() >= product(dims_) * sizeof(T) + offset_,
......@@ -198,7 +177,11 @@ class Tensor {
std::shared_ptr<Placeholder> holder_; // holds the memory block if allocated.
DDim dims_;
size_t offset_; // marks the begin of tensor data area.
// A PlaceHolder may be shared by more than one tensor. Some of them may be
// slices of the others. So the offset_ is introduced here to indicate the
// byte offset between PlaceHolder::ptr_ and where tensor's data really
// begins.
size_t offset_;
};
} // namespace framework
......
......@@ -19,7 +19,7 @@ TEST(Tensor, Dims) {
using namespace paddle::framework;
using namespace paddle::platform;
Tensor tt;
tt.set_dims(make_ddim({2, 3, 4}));
tt.Resize(make_ddim({2, 3, 4}));
DDim dims = tt.dims();
ASSERT_EQ(arity(dims), 3);
for (int i = 0; i < 3; ++i) {
......@@ -97,7 +97,7 @@ TEST(Tensor, MutableData) {
#endif
}
TEST(Tensor, ShareDataFrom) {
TEST(Tensor, ShareDataWith) {
using namespace paddle::framework;
using namespace paddle::platform;
{
......@@ -106,7 +106,7 @@ TEST(Tensor, ShareDataFrom) {
// Try to share data form uninitialized tensor
bool caught = false;
try {
dst_tensor.ShareDataFrom<float>(src_tensor);
dst_tensor.ShareDataWith<float>(src_tensor);
} catch (std::runtime_error& err) {
caught = true;
std::string msg =
......@@ -119,7 +119,7 @@ TEST(Tensor, ShareDataFrom) {
ASSERT_TRUE(caught);
src_tensor.mutable_data<int>(make_ddim({2, 3, 4}), CPUPlace());
dst_tensor.ShareDataFrom<int>(src_tensor);
dst_tensor.ShareDataWith<int>(src_tensor);
ASSERT_EQ(src_tensor.data<int>(), dst_tensor.data<int>());
}
......@@ -128,7 +128,7 @@ TEST(Tensor, ShareDataFrom) {
Tensor src_tensor;
Tensor dst_tensor;
src_tensor.mutable_data<int>(make_ddim({2, 3, 4}), GPUPlace());
dst_tensor.ShareDataFrom<int>(src_tensor);
dst_tensor.ShareDataWith<int>(src_tensor);
ASSERT_EQ(src_tensor.data<int>(), dst_tensor.data<int>());
}
#endif
......
......@@ -28,5 +28,17 @@ void Free(Place, void*);
template <class Place>
size_t Used(Place);
template <typename T, /* must be POD types */
typename Place /* platform::GPUPlace or platform::CPUPlace */,
typename std::enable_if<std::is_pod<T>::value>::type* = nullptr>
class PODDeleter {
public:
PODDeleter(Place place) : place_(place) {}
void operator()(T* ptr) { Free(place_, static_cast<void*>(ptr)); }
private:
Place place_;
};
} // namespace memory
} // namespace paddle
......@@ -31,7 +31,7 @@ protected:
"Inputs/Outputs of AddOp must all be set");
PADDLE_ENFORCE(inputs[0]->dims() == inputs[1]->dims(),
"Two input of Add Op's dimension must be same.");
outputs[0]->set_dims(inputs[0]->dims());
outputs[0]->Resize(inputs[0]->dims());
}
};
......
......@@ -35,7 +35,7 @@ protected:
PADDLE_ENFORCE(inputs[0]->dims().size() == 2, "X's dimension must be 2.");
PADDLE_ENFORCE(outputs[0]->dims().size() == 1,
"label's dimension must be 1.");
outputs[0]->set_dims(framework::make_ddim({inputs[0]->dims()[0]}));
outputs[0]->Resize(framework::make_ddim({inputs[0]->dims()[0]}));
}
};
......
......@@ -33,7 +33,7 @@ protected:
dim0[1] == dim1[0],
"First matrix's width must be equal with second matrix's height.");
PADDLE_ENFORCE(outputs.size() == 1, "The mul op must take one output");
outputs[0]->set_dims({dim0[0], dim1[1]});
outputs[0]->Resize({dim0[0], dim1[1]});
}
};
......
......@@ -30,7 +30,7 @@ protected:
PADDLE_ENFORCE(dim1.size() == 1, "The second input must be vector");
PADDLE_ENFORCE(dim0[1] == dim1[0], "The width of two input must be same");
PADDLE_ENFORCE(outputs.size() == 1, "The output size must be 1");
outputs[0]->set_dims(inputs[0]->dims());
outputs[0]->Resize(inputs[0]->dims());
}
};
......
......@@ -31,7 +31,7 @@ protected:
PADDLE_ENFORCE(outputs[0] != nullptr, "outputs[0] mast be set");
PADDLE_ENFORCE(inputs[0]->dims() == inputs[1]->dims(),
"Two input of SGD Op's dimension must be same.");
outputs[0]->set_dims(inputs[0]->dims());
outputs[0]->Resize(inputs[0]->dims());
}
};
......
......@@ -24,7 +24,7 @@ protected:
const std::vector<framework::Tensor *> &outputs) const override {
PADDLE_ENFORCE(inputs.size() == 1, "Sigmoid Op only have one input");
PADDLE_ENFORCE(outputs.size() == 1, "Sigmoid Op only have one output");
outputs[0]->set_dims(inputs[0]->dims());
outputs[0]->Resize(inputs[0]->dims());
}
};
......
......@@ -27,7 +27,7 @@ protected:
"The input of softmax op must be matrix");
PADDLE_ENFORCE(outputs.size() == 1, "Only one output is need for softmax");
outputs[0]->set_dims(inputs[0]->dims());
outputs[0]->Resize(inputs[0]->dims());
}
};
......
......@@ -46,7 +46,7 @@ PYBIND11_PLUGIN(core) {
[](const pd::Tensor& self) { return pd::vectorize(self.dims()); })
.def("set_dims",
[](pd::Tensor& self, const std::vector<int>& dim) {
self.set_dims(pd::make_ddim(dim));
self.Resize(pd::make_ddim(dim));
})
.def("alloc_float",
[](pd::Tensor& self) {
......
......@@ -86,7 +86,7 @@ void PyTensorSetFromArray(
dims.push_back((int)array.shape()[i]);
}
self.set_dims(framework::make_ddim(dims));
self.Resize(framework::make_ddim(dims));
auto *dst = self.mutable_data<T>(paddle::platform::CPUPlace());
std::memcpy(dst, array.data(), sizeof(T) * array.size());
}
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册
反馈
建议
客服 返回
顶部