提交 b8f4c859 编写于 作者: T typhoonzero

pserver runs in parallel

上级 128adf53
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/0.12.0 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 lite-v0.1
无相关合并请求
......@@ -24,6 +24,7 @@ limitations under the License. */
#include "paddle/fluid/framework/lod_tensor.h"
#include "paddle/fluid/framework/op_registry.h"
#include "paddle/fluid/framework/proto_desc.h"
#include "paddle/fluid/framework/threadpool.h"
#include "paddle/fluid/operators/detail/grpc_server.h"
#include "paddle/fluid/operators/detail/sendrecvop_utils.h"
#include "paddle/fluid/operators/detail/simple_block_queue.h"
......@@ -89,6 +90,7 @@ class ListenAndServOp : public framework::OperatorBase {
auto *block = Attr<framework::BlockDesc *>(kOptimizeBlock);
auto *program = block->Program();
int num_blocks = program->Size();
framework::Executor executor(dev_place);
// TODO(typhoonzero): change this to a while_op for every cluster-batch.
......@@ -132,12 +134,33 @@ class ListenAndServOp : public framework::OperatorBase {
rpc_service_->ShutDown();
break;
}
try {
executor.Run(*program, &recv_scope, block->ID(), /*global_block*/
false /*create_local_scope*/, false /*create_vars*/);
} catch (std::exception &e) {
LOG(ERROR) << "run sub program error " << e.what();
// put optimize blocks in the thread pool to start run, the last block
// should be global ops.
// NOTE: if is_gpu_place, CUDA kernels are laugched by multiple threads
// and this will still work.
std::vector<std::future<void>> fs;
for (int blkid = 0; blkid < num_blocks - 1; ++blkid) {
fs.push_back(framework::Async([&]() {
try {
executor.Run(*program, &recv_scope, blkid,
false /*create_local_scope*/, false /*create_vars*/);
} catch (std::exception &e) {
LOG(ERROR) << "run sub program error " << e.what();
}
}));
}
for (int blkid = 0; blkid < num_blocks - 1; ++blkid) fs[blkid].wait();
// Run global block at final step
if (num_blocks > 2) {
try {
executor.Run(*program, &recv_scope, num_blocks - 1,
false /*create_local_scope*/, false /*create_vars*/);
} catch (std::exception &e) {
LOG(ERROR) << "run sub program error " << e.what();
}
}
// Reset the received sparse variables, the sum operator would not
// sum the input sparse variables which rows is empty at the next
// mini-batch.
......
......@@ -307,15 +307,58 @@ class DistributeTranspiler:
# Iterate through the ops, and if an op and the optimize ops
# which located on current pserver are in one set, then
# append it into the sub program.
for _, op in enumerate(self.optimize_ops):
for _, opt_op in enumerate(opt_op_on_pserver):
if ufind.is_connected(op, opt_op):
if self._is_opt_op(op):
self._append_pserver_ops(optimize_block, op, endpoint,
default_main_program())
else:
self._append_pserver_non_opt_ops(optimize_block, op)
break
# We try to put optimization program run parallelly, assume
# optimization program always looks like:
#
# prevop -> prevop -> opt op -> following op -> following op; ->
# prevop -> prevop -> opt op -> following op -> following op; ->
# global op -> global op
#
# we put operators that can run parallelly to many program blocks.
# in above example, we seperate ops by the ";". Global ops must run
# after all the optimize ops finished.
global_ops = []
# HACK: optimization global ops only used to scale beta1 and beta2
# replace it with dependency engine.
for op in self.optimize_ops:
if op.type == "scale":
for in_name in op.input_arg_names:
if in_name.startswith("beta1_pow_acc") or\
in_name.startswith("beta2_pow_acc"):
global_ops.append(op)
print("##### global ops ", global_ops)
def __append_optimize_op__(op, block):
if self._is_opt_op(op):
self._append_pserver_ops(block, op, endpoint,
default_main_program())
else:
self._append_pserver_non_opt_ops(block, op)
# append op to the current block
per_opt_block = optimize_block
for _, opt_op in enumerate(opt_op_on_pserver):
for _, op in enumerate(self.optimize_ops):
# optimizer is connected to itself
if ufind.is_connected(op, opt_op) and \
op not in global_ops:
__append_optimize_op__(op, per_opt_block)
per_opt_block = pserver_program.create_block(0)
# append global ops
for glb_op in global_ops:
__append_optimize_op__(glb_op, per_opt_block)
# NOT USED: single block version:
#
# for _, op in enumerate(self.optimize_ops):
# for _, opt_op in enumerate(opt_op_on_pserver):
# if ufind.is_connected(op, opt_op):
# __append_optimize_op__(glb_op, optimize_block)
# break
# step5 append the listen_and_serv op
pserver_program.global_block().append_op(
type="listen_and_serv",
......@@ -660,10 +703,22 @@ class DistributeTranspiler:
# If one op's input is another op's output or
# one op's output is another op's input, we say
# the two operator is connected.
op1_input_names = op1.desc.input_arg_names()
def _append_inname_remove_beta(varname_list):
op_input_names = []
for in_name in varname_list:
# HACK: remove beta1 and beta2 to avoid let all
# ops connected.
if in_name.startswith("beta2_pow_acc") or \
in_name.startswith("beta1_pow_acc"):
continue
else:
op_input_names.append(in_name)
return op_input_names
op1_input_names = _append_inname_remove_beta(op1.desc.input_arg_names())
op1_output_names = op1.desc.output_arg_names()
op2_input_names = op2.desc.input_arg_names()
op2_input_names = _append_inname_remove_beta(op2.desc.input_arg_names())
op2_output_names = op2.desc.output_arg_names()
if set(op1_output_names) & set(op2_input_names) or \
......
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