提交 04b6c29e 编写于 作者: L lidanqing 提交者: Tao Luo

Improve mobilenetv2 INT8 performance by using INT8 relu as post-op (#17570)

* add INT8 conv+relu6 fuse and enbale mobilentv2 INT8 test
test=develop

* change fasle and 0.0 to fuse_brelu and brelu_threshold
test=develop

change the "fuse_relu||fuse_brelu" to "unsigned_output"
test=develop

* Use relu instead of brelu as INT8 post-op because INT8 brelu is not enabled in mkldnn v0.18
test=develop

* continuous-integration fix
test=develop
上级 6d8075ec
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_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 inplace_addto make_flag_adding_easier 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/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 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
无相关合并请求
......@@ -64,8 +64,10 @@ bool AnalysisPredictor::MkldnnQuantizer::CalculateScales() {
bool is_unsigned = false;
if (is_output && op->Type() == "conv2d") {
// output of conv2d with relu must be unsigned
is_unsigned = op->HasAttr("fuse_relu") &&
boost::get<bool>(op->GetAttr("fuse_relu"));
is_unsigned = (op->HasAttr("fuse_relu") &&
boost::get<bool>(op->GetAttr("fuse_relu"))) ||
(op->HasAttr("fuse_brelu") &&
boost::get<bool>(op->GetAttr("fuse_brelu")));
} else if (is_output && op->Type() == "relu") {
is_unsigned = true;
} else if (is_output &&
......
......@@ -176,6 +176,13 @@ if(WITH_MKLDNN)
endif()
inference_analysis_api_int8_test(test_analyzer_int8_mobilenet ${INT8_MOBILENET_MODEL_DIR} ${INT8_DATA_DIR} analyzer_int8_image_classification_tester.cc)
#mobilenetv2 int8
set(INT8_MOBILENETV2_MODEL_DIR "${INT8_DATA_DIR}/mobilenetv2")
if (NOT EXISTS ${INT8_MOBILENETV2_MODEL_DIR})
inference_download_and_uncompress(${INT8_MOBILENETV2_MODEL_DIR} "${INFERENCE_URL}/int8" "mobilenet_v2_int8_model.tar.gz" )
endif()
inference_analysis_api_int8_test(test_analyzer_int8_mobilenetv2 ${INT8_MOBILENETV2_MODEL_DIR} ${INT8_DATA_DIR} analyzer_int8_image_classification_tester.cc)
#resnet101 int8
set(INT8_RESNET101_MODEL_DIR "${INT8_DATA_DIR}/resnet101")
if (NOT EXISTS ${INT8_RESNET101_MODEL_DIR})
......
......@@ -288,6 +288,7 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
output->set_layout(DataLayout::kMKLDNN);
output->set_format(GetMKLDNNFormat(*dst_memory_p));
}
void ComputeINT8(const paddle::framework::ExecutionContext& ctx) const {
const bool is_test = ctx.Attr<bool>("is_test");
......@@ -325,7 +326,9 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
bool fuse_relu = ctx.Attr<bool>("fuse_relu");
bool fuse_residual_conn = ctx.Attr<bool>("fuse_residual_connection");
bool fuse_brelu = ctx.Attr<bool>("fuse_brelu");
float fuse_brelu_threshold = ctx.Attr<float>("fuse_brelu_threshold");
bool force_fp32_output = ctx.Attr<bool>("force_fp32_output");
bool unsigned_output = fuse_relu || fuse_brelu;
if (fuse_residual_conn) {
PADDLE_ENFORCE(force_fp32_output != true,
"residual fusion does not support force output with fp32");
......@@ -340,8 +343,6 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
"dilation in convolution is not implemented yet");
PADDLE_ENFORCE(is_conv3d != true, "int8 does not support conv3d currently");
PADDLE_ENFORCE(fuse_brelu != true,
"int8 does not support conv/relu6 fusion currently");
const T* input_data = input->data<T>();
......@@ -356,10 +357,11 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
mkldnn::memory::data_type src_dt =
paddle::framework::ToMKLDNNDataType(input->type());
auto dst_dt = (fuse_relu) ? paddle::framework::ToMKLDNNDataType(
framework::DataTypeTrait<uint8_t>::DataType)
: paddle::framework::ToMKLDNNDataType(
framework::DataTypeTrait<int8_t>::DataType);
auto dst_dt = unsigned_output
? paddle::framework::ToMKLDNNDataType(
framework::DataTypeTrait<uint8_t>::DataType)
: paddle::framework::ToMKLDNNDataType(
framework::DataTypeTrait<int8_t>::DataType);
if (force_fp32_output) {
dst_dt = paddle::framework::ToMKLDNNDataType(
......@@ -377,13 +379,12 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
key.reserve(MaxKeyLength);
platform::ConvMKLDNNHandler::AppendKey(
&key, src_tz, weights_tz, strides, paddings, dilations, groups, src_dt,
input->format(), fuse_relu, fuse_residual_conn, false /*fuse_brelu*/,
input->format(), fuse_relu, fuse_residual_conn, fuse_brelu,
ctx.op().Input("Input") + ctx.op().Input("Filter"));
const std::string key_conv_pd = key + "@conv_pd";
bool need_s8_to_u8 = false;
std::shared_ptr<mkldnn::convolution_forward> conv_p = nullptr;
std::shared_ptr<mkldnn::memory> src_memory_p = nullptr;
std::shared_ptr<mkldnn::memory> user_src_memory_p = nullptr;
......@@ -456,6 +457,9 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
platform::MKLDNNMemDesc(dst_tz, dst_dt, chosen_memory_format);
// create a conv primitive descriptor and save it for usage in backward
// TODO(lidanqing): We use relu post-op instead of brelu post-op cause
// mkldnn v0.18 does not support INT8 brelu post-op. Use code in /**/ when
// v0.20 is enabled
if (bias) {
bias_tz = paddle::framework::vectorize2int(bias->dims());
auto bias_md = platform::MKLDNNMemDesc(bias_tz, memory::data_type::s32,
......@@ -463,16 +467,16 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
conv_pd = ConvFwdPrimitiveDesc(
src_md, weights_md, bias_md, dst_md, strides, paddings,
mkldnn_engine, fuse_relu, fuse_residual_conn, false /*fuse_brelu*/,
0.0 /*fuse_brelu_threshold*/, output_shift_scale, sum_scale,
is_test);
mkldnn_engine, fuse_relu || fuse_brelu /*fuse_relu*/,
fuse_residual_conn, false /*fuse_brelu*/, fuse_brelu_threshold,
output_shift_scale, sum_scale, is_test);
} else {
conv_pd = ConvFwdPrimitiveDesc(src_md, weights_md, dst_md, strides,
paddings, mkldnn_engine, fuse_relu,
fuse_residual_conn, false /*fuse_brelu*/,
0.0 /*fuse_brelu_threshold*/,
output_shift_scale, sum_scale, is_test);
conv_pd = ConvFwdPrimitiveDesc(
src_md, weights_md, dst_md, strides, paddings, mkldnn_engine,
fuse_relu || fuse_brelu /*fuse_relu*/, fuse_residual_conn,
false /*fuse_brelu*/, fuse_brelu_threshold, output_shift_scale,
sum_scale, is_test);
}
// Save conv_pd/src_memory/weights_memory for backward pass
dev_ctx.SetBlob(key_conv_pd, conv_pd);
......@@ -514,7 +518,7 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
ctx, output, residual_param, user_residual_md, handler,
&pipeline);
} else {
need_s8_to_u8 = fuse_relu;
need_s8_to_u8 = unsigned_output;
dst_memory_p = platform::SetDstMemory<int8_t>(
ctx, output, residual_param, user_residual_md, handler,
&pipeline);
......@@ -525,12 +529,12 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
dst_memory_p =
platform::SetDstMemory<uint8_t>(ctx, output, handler);
} else {
need_s8_to_u8 = fuse_relu;
need_s8_to_u8 = unsigned_output;
dst_memory_p = platform::SetDstMemory<int8_t>(ctx, output, handler);
}
}
} else if (!force_fp32_output) {
if (fuse_relu) {
if (unsigned_output) {
dst_memory_p = platform::SetDstMemory<uint8_t>(ctx, output, handler);
} else {
dst_memory_p = platform::SetDstMemory<int8_t>(ctx, output, handler);
......@@ -602,12 +606,12 @@ class ConvMKLDNNOpKernel : public paddle::framework::OpKernel<T> {
platform::SetDstMemoryHandler<uint8_t>(ctx, output, handler,
&dst_memory_p);
} else {
need_s8_to_u8 = fuse_relu;
need_s8_to_u8 = unsigned_output;
platform::SetDstMemoryHandler<int8_t>(ctx, output, handler,
&dst_memory_p);
}
} else if (!force_fp32_output) {
if (fuse_relu) {
if (unsigned_output) {
platform::SetDstMemoryHandler<uint8_t>(ctx, output, handler,
&dst_memory_p);
} else {
......
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