提交 1203ebc4 编写于 作者: T tensor-tang

add mkldnn fc backward

上级 90d5be74
develop 1.8.5 2.0.1-rocm-post 2.4.1 Ligoml-patch-1 OliverLPH-patch-1 OliverLPH-patch-2 PaddlePM-patch-1 PaddlePM-patch-2 ZHUI-patch-1 add_default_att add_kylinv10 add_model_benchmark_ci add_some_yaml_config addfile all_new_design_exec ascendrc ascendrelease bugfix-eval-frame-leakgae cherry-pick-fix-customOP-random-fail cherry_undefined_var compile_windows cp_2.4_fix_numpy 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-run-program-grad-node-mem fix_check fix_concat_slice fix_custom_device_copy_sync fix_dataloader_memory_leak fix_dlpack_for fix_imperative_dygraph_error fix_newexe_gc fix_npu_ci fix_op_flops fix_retry_ci fix_rnn_docs fix_tensor_type fix_undefined_var fix_var_stop_gradient_error 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 hack_event improve_sccache incuabte/new_frl incubate/frl_train_eval incubate/infrt incubate/lite incubate/new_frl incubate/new_frl_rc incubate/stride inplace_addto layer_norm make_flag_adding_easier master matmul_double_grad 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 operator_opt paddle_tiny_install paralleltest pass-compile-eval-frame preln_ernie prv-disable-more-cache prv-md-even-more prv-onednn-2.5 prv-reshape-mkldnn-ut2 pten_tensor_refactor release-deleted/2.5 release-rc/2.5 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/2.5 release/lite-0.1 release/llm_2.5 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 revert-47325-remove_cudnn_hardcode revert-47645-add_npu_storage_dims revert-48815-set_free_when_no_cache_hit_default_value_true revert-49499-test_ninja_on_ci revert-49654-prim_api_gen revert-49673-modify_get_single_cov revert-49763-fix_static_composite_gen revert-50158-fix_found_inf_bug_for_custom_optimizer revert-50188-refine_optimizer_create_accumulators revert-50335-fix_optminizer_set_auxiliary_var_bug revert-51676-flag_delete revert-51850-fix_softmaxce_dev revert-52175-dev_peak_memory revert-52186-deve revert-52523-test_py38 revert-52912-develop revert-53248-set_cmake_policy revert-54029-fix_windows_compile_bug revert-54068-support_translating_op_attribute revert-54214-modify_cmake_dependencies revert-54370-offline_pslib revert-54391-fix_cmake_md5error revert-54411-fix_cpp17_compile revert-54466-offline_pslib revert-54480-cmake-rocksdb revert-55568-fix_BF16_bug1 revert-56328-new_ir_support_vector_type_place_transfer revert-56366-fix_openssl_bug revert-56545-revert-56366-fix_openssl_bug revert-56620-fix_new_ir_ocr_bug revert-56925-check_inputs_grad_semantic revert-57005-refine_stride_flag rocm_dev_0217 sd_conv_linear_autocast semi-auto/rule-base support-0D-sort support_weight_transpose test_benchmark_ci test_feature_precision_test_c test_for_Filtetfiles test_model_benchmark test_model_benchmark_ci zhiqiu-patch-1 v2.5.1 v2.5.0 v2.5.0-rc1 v2.5.0-rc0 v2.4.2 v2.4.1 v2.4.0 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
7 合并请求!11636[IMPORTANT] MKLDNN layout: Support for sum operator,!8482Release/0.11.0,!8190Release/0.11.0,!8189Release/0.11.0,!6633给线性回归的get-started代码加上了预测的示例~~,!4615Feature/tensor array add python binding,!3337Enable mkldnn_fc for general format
......@@ -77,7 +77,6 @@ void MkldnnFcLayer::reshape() {
void MkldnnFcLayer::forward(PassType passType) {
Layer::forward(passType);
reshape();
{
......@@ -97,6 +96,40 @@ void MkldnnFcLayer::forward(PassType passType) {
}
void MkldnnFcLayer::backward(const UpdateCallback& callback) {
; // bool hasBias = biases_ && biases_->getWGrad();
/* Do derivation */ {
REGISTER_TIMER_INFO("BpActTimer", getName().c_str());
backwardActivation();
}
bool hasBias = biases_ && biases_->getWGrad();
{
REGISTER_TIMER_INFO("mkldnn_bwdTimer", getName().c_str());
real* inVal = getInputValue(0)->getData();
real* inGrad =
getInputGrad(0) != nullptr ? getInputGrad(0)->getData() : NULL;
real* outGrad = getOutputGrad()->getData();
real* wgtGrad = weight_->getWGrad()->getData();
real* wgtVal = weight_->getW()->getData();
real* biasGrad = hasBias ? biases_->getWGrad()->getData() : NULL;
mkldnnBackwardFC(bs_,
ic_,
ih_,
iw_,
inGrad,
inVal,
oc_,
outGrad,
wgtGrad,
wgtVal,
biasGrad);
}
{
REGISTER_TIMER_INFO("WeightUpdate", getName().c_str());
weight_->getParameterPtr()->incUpdate(callback);
if (hasBias) {
biases_->getParameterPtr()->incUpdate(callback);
}
}
}
} // namespace paddle
......@@ -88,6 +88,94 @@ void MkldnnLayer::mkldnnForwardFC(int bs,
stream_->submit(pipelineFwd_);
}
void MkldnnLayer::resetBackwardFC(int bs,
int ic,
int ih,
int iw,
real* botDiff,
real* botData,
int oc,
real* topDiff,
real* wgtDiff,
real* wgtData,
real* biasDiff) {
bool hasSpatial = ih == 1 && iw == 1 ? false : true;
engine_ = CpuEngine::Instance().getEngine();
// backward weight
mem::desc botMD = hasSpatial ? createMD({bs, ic, ih, iw}, format::nchw)
: createMD({bs, ic}, format::nc);
mem::desc wgtMD = hasSpatial ? createMD({oc, ic, ih, iw}, format::oihw)
: createMD({oc, ic}, format::oi);
mem::desc topMD = createMD({bs, oc}, format::nc);
mem::desc biasMD = biasDiff != NULL ? createMD({oc}, format::x)
: createMD({}, format::format_undef);
fc_fwd::desc fwdDesc =
fc_fwd::desc(mkldnn::prop_kind::forward, botMD, wgtMD, topMD);
fc_fwd::primitive_desc fwdPD = fc_fwd::primitive_desc(fwdDesc, engine_);
fc_bwdWgt::desc bwdWgtDesc =
biasDiff != NULL ? fc_bwdWgt::desc(botMD, wgtMD, biasMD, topMD)
: fc_bwdWgt::desc(botMD, wgtMD, topMD);
fc_bwdWgt::primitive_desc bwdWgtPD =
fc_bwdWgt::primitive_desc(bwdWgtDesc, engine_, fwdPD);
mem botVal = mem(mem::primitive_desc(botMD, engine_), botData);
mem wgtGrad = mem(mem::primitive_desc(wgtMD, engine_), wgtDiff);
mem topGrad = mem(mem::primitive_desc(topMD, engine_), topDiff);
if (biasDiff != NULL) {
mem biasGrad = mem(mem::primitive_desc(biasMD, engine_), biasDiff);
bwdWgt_.reset(new fc_bwdWgt(bwdWgtPD, botVal, topGrad, wgtGrad, biasGrad));
} else {
bwdWgt_.reset(new fc_bwdWgt(bwdWgtPD, botVal, topGrad, wgtGrad));
}
pipelineBwd_.clear();
pipelineBwd_.push_back(*bwdWgt_);
// backward data
if (botDiff == NULL) {
return;
}
fc_bwdData::desc bwdDataDesc = fc_bwdData::desc(botMD, wgtMD, topMD);
fc_bwdData::primitive_desc bwdDataPD =
fc_bwdData::primitive_desc(bwdDataDesc, engine_, fwdPD);
mem botGrad = mem(mem::primitive_desc(botMD, engine_), botDiff);
mem wgtVal = mem(mem::primitive_desc(wgtMD, engine_), wgtData);
bwdData_.reset(new fc_bwdData(bwdDataPD, topGrad, wgtVal, botGrad));
pipelineBwd_.push_back(*bwdData_);
}
void MkldnnLayer::mkldnnBackwardFC(int bs,
int ic,
int ih,
int iw,
real* botDiff,
real* botData,
int oc,
real* topDiff,
real* wgtDiff,
real* wgtData,
real* biasDiff) {
// if input size changed, reset it
resetBackwardFC(bs,
ic,
ih,
iw,
botDiff,
botData,
oc,
topDiff,
wgtDiff,
wgtData,
biasDiff);
// just forward
// update botdata
stream_->submit(pipelineBwd_);
}
mem::desc MkldnnLayer::createMD(mem::dims dims,
mem::format fmt,
mem::data_type type) {
......
......@@ -42,6 +42,8 @@ protected:
std::shared_ptr<MkldnnStream> stream_;
std::shared_ptr<mkldnn::primitive> fwd_;
std::shared_ptr<mkldnn::primitive> bwdWgt_;
std::shared_ptr<mkldnn::primitive> bwdData_;
std::vector<mkldnn::primitive> pipelineFwd_;
std::vector<mkldnn::primitive> pipelineBwd_;
......@@ -56,7 +58,10 @@ public:
oh_(0),
ow_(0),
engine_(mkldnn::engine::cpu, 0),
stream_(nullptr) {}
stream_(nullptr),
fwd_(nullptr),
bwdWgt_(nullptr),
bwdData_(nullptr) {}
~MkldnnLayer() {}
......@@ -82,6 +87,30 @@ public:
real* wgtData,
real* biasData);
void resetBackwardFC(int bs,
int ic,
int ih,
int iw,
real* botDiff,
real* botData,
int oc,
real* topDiff,
real* wgtDiff,
real* wgtData,
real* biasDiff);
void mkldnnBackwardFC(int bs,
int ic,
int ih,
int iw,
real* botDiff,
real* botData,
int oc,
real* topDiff,
real* wgtDiff,
real* wgtData,
real* biasDiff);
// TODO(TJ): move to MkldnnMatrix
// create memory desc
inline mkldnn::memory::desc createMD(
......
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