未验证 提交 0b8067c0 编写于 作者: W Wu Yi 提交者: GitHub

fix dist train reduce mode (#13068)

* fix dist train reduce mode

* fix previous fix
上级 823c4f87
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 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/1.0.0 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 lite-v0.1
无相关合并请求
......@@ -744,7 +744,7 @@ void MultiDevSSAGraphBuilder::CreateDistTrainOp(ir::Graph *result,
.emplace(varname, op_dev_id);
}
} else {
PADDLE_ENFORCE(
PADDLE_THROW(
"the distribute training related op should be in [split_byref, "
"concat].");
}
......
......@@ -82,8 +82,18 @@ class TestDistRunnerBase(object):
strategy = fluid.ExecutionStrategy()
strategy.num_threads = 1
strategy.allow_op_delay = False
build_stra = fluid.BuildStrategy()
if args.use_reduce:
build_stra.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.Reduce
else:
build_stra.reduce_strategy = fluid.BuildStrategy.ReduceStrategy.AllReduce
exe = fluid.ParallelExecutor(
True, loss_name=avg_cost.name, exec_strategy=strategy)
True,
loss_name=avg_cost.name,
exec_strategy=strategy,
build_strategy=build_stra)
feed_var_list = [
var for var in trainer_prog.global_block().vars.values()
......@@ -123,6 +133,7 @@ def runtime_main(test_class):
'--current_endpoint', type=str, required=False, default="")
parser.add_argument('--sync_mode', action='store_true')
parser.add_argument('--mem_opt', action='store_true')
parser.add_argument('--use_reduce', action='store_true')
args = parser.parse_args()
......@@ -149,20 +160,25 @@ class TestDistBase(unittest.TestCase):
self._python_interp = "python"
self._sync_mode = True
self._mem_opt = False
self._use_reduce = False
self._setup_config()
def start_pserver(self, model_file, check_error_log):
ps0_ep, ps1_ep = self._ps_endpoints.split(",")
ps_cmd = "%s %s --role pserver --endpoints %s --trainer_id 0 --current_endpoint %s --trainers %d --is_dist %s %s"
sync_mode_str = "--sync_mode" if self._sync_mode else ""
mem_opt_str = "--mem_opt" if self._mem_opt else ""
ps_cmd = "%s %s --role pserver --endpoints %s --trainer_id 0 --current_endpoint %s --trainers %d --is_dist"
ps0_cmd = ps_cmd % \
(self._python_interp, model_file, self._ps_endpoints, ps0_ep,
self._trainers, sync_mode_str, mem_opt_str)
self._trainers)
ps1_cmd = ps_cmd % \
(self._python_interp, model_file, self._ps_endpoints, ps1_ep,
self._trainers, sync_mode_str, mem_opt_str)
self._trainers)
if self._sync_mode:
ps0_cmd += " --sync_mode"
ps1_cmd += " --sync_mode"
if self._mem_opt:
ps0_cmd += " --mem_opt"
ps1_cmd += " --mem_opt"
ps0_pipe = subprocess.PIPE
ps1_pipe = subprocess.PIPE
......@@ -242,17 +258,23 @@ class TestDistBase(unittest.TestCase):
self._wait_ps_ready(ps1.pid)
ps0_ep, ps1_ep = self._ps_endpoints.split(",")
tr_cmd = "%s %s --role trainer --endpoints %s --trainer_id %d --current_endpoint %s --trainers %d --is_dist %s %s"
sync_mode_str = "--sync_mode" if self._sync_mode else ""
mem_opt_str = "--mem_opt" if self._mem_opt else ""
tr_cmd = "%s %s --role trainer --endpoints %s --trainer_id %d --current_endpoint %s --trainers %d --is_dist"
tr0_cmd = tr_cmd % \
(self._python_interp, model_file, self._ps_endpoints,
0, ps0_ep,
self._trainers, sync_mode_str, mem_opt_str)
0, ps0_ep, self._trainers)
tr1_cmd = tr_cmd % \
(self._python_interp, model_file, self._ps_endpoints,
1, ps1_ep,
self._trainers, sync_mode_str, mem_opt_str)
1, ps1_ep, self._trainers)
if self._sync_mode:
tr0_cmd += " --sync_mode"
tr1_cmd += " --sync_mode"
if self._mem_opt:
tr0_cmd += " --mem_opt"
tr1_cmd += " --mem_opt"
if self._use_reduce:
tr0_cmd += " --use_reduce"
tr1_cmd += " --use_reduce"
env0 = {"CUDA_VISIBLE_DEVICES": "0"}
env1 = {"CUDA_VISIBLE_DEVICES": "1"}
......@@ -303,6 +325,8 @@ class TestDistBase(unittest.TestCase):
# FIXME: use terminate() instead of sigkill.
os.kill(ps0.pid, signal.SIGKILL)
os.kill(ps1.pid, signal.SIGKILL)
ps0.terminate()
ps1.terminate()
ps0.wait()
ps1.wait()
FNULL.close()
......
......@@ -20,6 +20,7 @@ from test_dist_base import TestDistBase
class TestDistMnist2x2(TestDistBase):
def _setup_config(self):
self._sync_mode = True
self._use_reduce = False
def test_se_resnext(self):
self.check_with_place("dist_mnist.py", delta=1e-7)
......@@ -37,10 +38,30 @@ class TestDistMnist2x2WithMemopt(TestDistBase):
class TestDistMnistAsync(TestDistBase):
def _setup_config(self):
self._sync_mode = False
self._use_reduce = False
def test_se_resnext(self):
self.check_with_place("dist_mnist.py", delta=200)
# FIXME(typhoonzero): enable these tests once we have 4
# 4 GPUs on CI machine, and the base class should be updated.
#
# class TestDistMnist2x2ReduceMode(TestDistBase):
# def _setup_config(self):
# self._sync_mode = True
# self._use_reduce = True
# def test_se_resnext(self):
# self.check_with_place("dist_mnist.py", delta=1e-7)
# class TestDistMnistAsyncReduceMode(TestDistBase):
# def _setup_config(self):
# self._sync_mode = False
# self._use_reduce = True
# def test_se_resnext(self):
# self.check_with_place("dist_mnist.py", delta=200)
if __name__ == "__main__":
unittest.main()
......@@ -273,6 +273,10 @@ class DistributeTranspiler(object):
name=framework.generate_control_dev_var_name())
grad_name_to_send_dummy_out[grad_varname] = dummy_output
# get send op_role_var, if not splited, the grad should have .trainer suffix
# if splited, grad should be the original grad var name (split_by_ref and send
# will be on the same place). ParallelExecutor
# will use op_role_var to get expected device place to run this op.
program.global_block()._insert_op(
index=index + 1,
type="send",
......@@ -281,8 +285,10 @@ class DistributeTranspiler(object):
attrs={
"epmap": eplist,
RPC_OP_ROLE_ATTR_NAME: RPC_OP_ROLE_ATTR_VALUE,
OP_ROLE_VAR_ATTR_NAME:
[self.grad_name_to_param_name[grad_varname], grad_varname],
OP_ROLE_VAR_ATTR_NAME: [
self.grad_name_to_param_name[grad_varname],
splited_grad_varname
],
"sync_mode": not self.sync_mode,
})
for _, var in enumerate(splited_vars):
......@@ -326,6 +332,15 @@ class DistributeTranspiler(object):
recv_dep_in = grad_name_to_send_dummy_out[
self.param_name_to_grad_name[param_varname]]
all_recv_outputs.extend(splited_var)
# get recv op_role_var, if not splited, the grad should have .trainer suffix
# if splited, grad should be the original grad var name. ParallelExecutor
# will use op_role_var to get expected device place to run this op.
orig_grad_name = self.param_name_to_grad_name[param_varname]
recv_op_role_var_name = orig_grad_name
splited_trainer_grad = self.grad_var_mapping[orig_grad_name]
if len(splited_trainer_grad) == 1:
recv_op_role_var_name = splited_trainer_grad[0].name
program.global_block().append_op(
type="recv",
inputs={"X": [recv_dep_in]},
......@@ -333,10 +348,8 @@ class DistributeTranspiler(object):
attrs={
"epmap": eps,
RPC_OP_ROLE_ATTR_NAME: RPC_OP_ROLE_ATTR_VALUE,
OP_ROLE_VAR_ATTR_NAME: [
param_varname,
self.param_name_to_grad_name[param_varname]
],
OP_ROLE_VAR_ATTR_NAME:
[param_varname, recv_op_role_var_name],
"sync_mode": not self.sync_mode
})
......
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