William Falcon
GitHub
parent
b01ad75700
commit
a0c4365278
|
|
@ -3,3 +3,5 @@ from pytorch_lightning.accelerator_backends.tpu_backend import TPUBackend
|
|||
from pytorch_lightning.accelerator_backends.dp_backend import DataParallelBackend
|
||||
from pytorch_lightning.accelerator_backends.ddp_spawn_backend import DDPSpawnBackend
|
||||
from pytorch_lightning.accelerator_backends.cpu_backend import CPUBackend
|
||||
from pytorch_lightning.accelerator_backends.ddp_backend import DDPBackend
|
||||
from pytorch_lightning.accelerator_backends.ddp2_backend import DDP2Backend
|
||||
|
|
|
|||
|
|
@ -0,0 +1,160 @@
|
|||
# Copyright The PyTorch Lightning team.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License
|
||||
|
||||
import os
|
||||
import torch
|
||||
from pytorch_lightning.utilities import NATIVE_AMP_AVALAIBLE
|
||||
from pytorch_lightning.utilities.distributed import rank_zero_only
|
||||
from pytorch_lightning import _logger as log
|
||||
from pytorch_lightning.utilities.exceptions import MisconfigurationException
|
||||
|
||||
try:
|
||||
from hydra.utils import to_absolute_path, get_original_cwd
|
||||
from hydra.core.hydra_config import HydraConfig
|
||||
except ImportError:
|
||||
HYDRA_AVAILABLE = False
|
||||
else:
|
||||
HYDRA_AVAILABLE = True
|
||||
|
||||
try:
|
||||
from apex import amp
|
||||
except ImportError:
|
||||
APEX_AVAILABLE = False
|
||||
else:
|
||||
APEX_AVAILABLE = True
|
||||
|
||||
|
||||
class DDP2Backend(object):
|
||||
|
||||
def __init__(self, trainer):
|
||||
self.trainer = trainer
|
||||
self.task_idx = None
|
||||
|
||||
def setup(self):
|
||||
self._resolve_task_idx()
|
||||
|
||||
def _resolve_task_idx(self):
|
||||
if self.trainer.is_slurm_managing_tasks:
|
||||
self.task_idx = int(os.environ['SLURM_LOCALID'])
|
||||
else:
|
||||
# torchelastic or general non_slurm ddp2
|
||||
try:
|
||||
self.task_idx = int(os.environ['LOCAL_RANK'])
|
||||
except Exception as e:
|
||||
m = 'ddp2 only works in SLURM or via torchelastic with the WORLD_SIZE, LOCAL_RANK, GROUP_RANK flags'
|
||||
raise MisconfigurationException(m)
|
||||
|
||||
def train(self, model):
|
||||
self.ddp_train(process_idx=self.task_idx, mp_queue=None, model=model)
|
||||
|
||||
def ddp_train(self, process_idx, mp_queue, model, is_master=False, proc_offset=0):
|
||||
"""
|
||||
Entry point for ddp
|
||||
|
||||
Args:
|
||||
process_idx:
|
||||
mp_queue: multiprocessing queue
|
||||
model:
|
||||
is_master:
|
||||
proc_offset:
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
# offset the process id if requested
|
||||
process_idx = process_idx + proc_offset
|
||||
|
||||
# show progressbar only on progress_rank 0
|
||||
if (self.trainer.node_rank != 0 or process_idx != 0) and self.trainer.progress_bar_callback is not None:
|
||||
self.trainer.progress_bar_callback.disable()
|
||||
|
||||
self.trainer.local_rank = self.trainer.node_rank
|
||||
self.trainer.global_rank = self.trainer.node_rank
|
||||
self.trainer.world_size = self.trainer.num_nodes
|
||||
|
||||
# set warning rank
|
||||
rank_zero_only.rank = self.trainer.global_rank
|
||||
|
||||
# set up server using proc 0's ip address
|
||||
# try to init for 20 times at max in case ports are taken
|
||||
# where to store ip_table
|
||||
model.trainer = self.trainer
|
||||
model.init_ddp_connection(
|
||||
self.trainer.global_rank,
|
||||
self.trainer.world_size,
|
||||
self.trainer.is_slurm_managing_tasks
|
||||
)
|
||||
|
||||
# call setup after the ddp process has connected
|
||||
self.trainer.call_setup_hook(model)
|
||||
|
||||
# on world_size=0 let everyone know training is starting
|
||||
if self.trainer.is_global_zero:
|
||||
log.info('-' * 100)
|
||||
log.info(f'distributed_backend={self.trainer.distributed_backend}')
|
||||
log.info(f'All DDP processes registered. Starting ddp with {self.trainer.world_size} processes')
|
||||
log.info('-' * 100)
|
||||
|
||||
# CHOOSE OPTIMIZER
|
||||
# allow for lr schedulers as well
|
||||
optimizers, lr_schedulers, optimizer_frequencies = self.trainer.init_optimizers(model)
|
||||
self.trainer.optimizers = optimizers
|
||||
self.trainer.lr_schedulers = lr_schedulers
|
||||
self.trainer.optimizer_frequencies = optimizer_frequencies
|
||||
|
||||
# MODEL
|
||||
# copy model to each gpu
|
||||
if self.trainer.on_gpu:
|
||||
gpu_idx = process_idx
|
||||
|
||||
# when using ddp, the master process (proc 0) continues running as the main one
|
||||
# this means that the local rank will always be 0
|
||||
# (even if cuda visible devices has other visible gpus)
|
||||
# this means that the master process needs to pull the 0th visible index as the device number
|
||||
if is_master:
|
||||
available_gpus = os.environ['CUDA_VISIBLE_DEVICES'].split(',')
|
||||
gpu_idx = int(available_gpus[self.trainer.local_rank])
|
||||
|
||||
self.trainer.root_gpu = gpu_idx
|
||||
torch.cuda.set_device(self.trainer.root_gpu)
|
||||
model.cuda(self.trainer.root_gpu)
|
||||
|
||||
# set model properties before going into wrapper
|
||||
self.trainer.copy_trainer_model_properties(model)
|
||||
|
||||
# AMP
|
||||
# run through amp wrapper before going to distributed DP
|
||||
# TODO: remove with dropping NVIDIA AMP support
|
||||
if self.trainer.use_amp and not NATIVE_AMP_AVALAIBLE:
|
||||
model, optimizers = model.configure_apex(amp, model, self.trainer.optimizers, self.trainer.amp_level)
|
||||
self.trainer.optimizers = optimizers
|
||||
self.trainer.reinit_scheduler_properties(self.trainer.optimizers, self.trainer.lr_schedulers)
|
||||
|
||||
# DDP2 uses all GPUs on the machine
|
||||
device_ids = self.trainer.data_parallel_device_ids
|
||||
|
||||
# allow user to configure ddp
|
||||
model = model.configure_ddp(model, device_ids)
|
||||
|
||||
# continue training routine
|
||||
results = self.trainer.run_pretrain_routine(model)
|
||||
|
||||
# get original model
|
||||
model = self.trainer.get_model()
|
||||
|
||||
# persist info in ddp_spawn
|
||||
self.trainer.transfer_distrib_spawn_state_on_fit_end(model, mp_queue, results)
|
||||
|
||||
# clean up memory
|
||||
torch.cuda.empty_cache()
|
||||
|
|
@ -0,0 +1,229 @@
|
|||
# Copyright The PyTorch Lightning team.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License
|
||||
|
||||
import os
|
||||
import torch
|
||||
import subprocess
|
||||
import sys
|
||||
from time import sleep
|
||||
import numpy as np
|
||||
from os.path import abspath
|
||||
from pytorch_lightning.utilities import NATIVE_AMP_AVALAIBLE
|
||||
from pytorch_lightning.utilities.distributed import rank_zero_only
|
||||
from pytorch_lightning import _logger as log
|
||||
from typing import Optional
|
||||
|
||||
try:
|
||||
from hydra.utils import to_absolute_path, get_original_cwd
|
||||
from hydra.core.hydra_config import HydraConfig
|
||||
except ImportError:
|
||||
HYDRA_AVAILABLE = False
|
||||
else:
|
||||
HYDRA_AVAILABLE = True
|
||||
|
||||
try:
|
||||
from apex import amp
|
||||
except ImportError:
|
||||
APEX_AVAILABLE = False
|
||||
else:
|
||||
APEX_AVAILABLE = True
|
||||
|
||||
|
||||
class DDPBackend(object):
|
||||
|
||||
def __init__(self, trainer):
|
||||
self.trainer = trainer
|
||||
self.task_idx = None
|
||||
|
||||
def slurm_setup(self):
|
||||
self.task_idx = int(os.environ['SLURM_LOCALID'])
|
||||
|
||||
def torchelastic_setup(self):
|
||||
self.task_idx = int(os.environ['LOCAL_RANK'])
|
||||
|
||||
def train(self, model):
|
||||
self.ddp_train(process_idx=self.task_idx, mp_queue=None, model=model)
|
||||
|
||||
def spawn_ddp_children(self, model):
|
||||
port = os.environ['MASTER_PORT']
|
||||
|
||||
master_address = '127.0.0.1' if 'MASTER_ADDR' not in os.environ else os.environ['MASTER_ADDR']
|
||||
os.environ['MASTER_PORT'] = f'{port}'
|
||||
os.environ['MASTER_ADDR'] = f'{master_address}'
|
||||
|
||||
# allow the user to pass the node rank
|
||||
node_rank = '0'
|
||||
if 'NODE_RANK' in os.environ:
|
||||
node_rank = os.environ['NODE_RANK']
|
||||
if 'GROUP_RANK' in os.environ:
|
||||
node_rank = os.environ['GROUP_RANK']
|
||||
|
||||
os.environ['NODE_RANK'] = node_rank
|
||||
os.environ['LOCAL_RANK'] = '0'
|
||||
|
||||
# when user is using hydra find the absolute path
|
||||
path_lib = abspath if not HYDRA_AVAILABLE else to_absolute_path
|
||||
|
||||
# pull out the commands used to run the script and resolve the abs file path
|
||||
command = sys.argv
|
||||
try:
|
||||
full_path = path_lib(command[0])
|
||||
except Exception as e:
|
||||
full_path = abspath(command[0])
|
||||
|
||||
command[0] = full_path
|
||||
# use the same python interpreter and actually running
|
||||
command = [sys.executable] + command
|
||||
|
||||
# since this script sets the visible devices we replace the gpus flag with a number
|
||||
num_gpus = os.environ['CUDA_VISIBLE_DEVICES'].split(',').__len__()
|
||||
|
||||
if '--gpus' in command:
|
||||
gpu_flag_idx = command.index('--gpus')
|
||||
command[gpu_flag_idx + 1] = f'{num_gpus}'
|
||||
|
||||
os.environ['WORLD_SIZE'] = f'{num_gpus * self.trainer.num_nodes}'
|
||||
|
||||
self.trainer.interactive_ddp_procs = []
|
||||
for local_rank in range(1, self.trainer.num_processes):
|
||||
env_copy = os.environ.copy()
|
||||
env_copy['LOCAL_RANK'] = f'{local_rank}'
|
||||
|
||||
# start process
|
||||
# if hydra is available and initialized, make sure to set the cwd correctly
|
||||
cwd: Optional[str] = None
|
||||
if HYDRA_AVAILABLE:
|
||||
if HydraConfig.initialized():
|
||||
cwd = get_original_cwd()
|
||||
proc = subprocess.Popen(command, env=env_copy, cwd=cwd)
|
||||
self.trainer.interactive_ddp_procs.append(proc)
|
||||
|
||||
# starting all processes at once can cause issues
|
||||
# with dataloaders delay between 1-10 seconds
|
||||
delay = np.random.uniform(1, 5, 1)[0]
|
||||
sleep(delay)
|
||||
|
||||
local_rank = 0
|
||||
results = self.ddp_train(local_rank, mp_queue=None, model=model, is_master=True)
|
||||
del os.environ['WORLD_SIZE']
|
||||
|
||||
return results
|
||||
|
||||
def ddp_train(self, process_idx, mp_queue, model, is_master=False, proc_offset=0):
|
||||
"""
|
||||
Entry point for ddp
|
||||
|
||||
Args:
|
||||
process_idx:
|
||||
mp_queue: multiprocessing queue
|
||||
model:
|
||||
is_master:
|
||||
proc_offset:
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
# offset the process id if requested
|
||||
process_idx = process_idx + proc_offset
|
||||
|
||||
# show progressbar only on progress_rank 0
|
||||
if (self.trainer.node_rank != 0 or process_idx != 0) and self.trainer.progress_bar_callback is not None:
|
||||
self.trainer.progress_bar_callback.disable()
|
||||
|
||||
# determine which process we are and world size
|
||||
self.trainer.local_rank = process_idx
|
||||
self.trainer.global_rank = self.trainer.node_rank * self.trainer.num_processes + process_idx
|
||||
self.trainer.world_size = self.trainer.num_nodes * self.trainer.num_processes
|
||||
|
||||
# set warning rank
|
||||
rank_zero_only.rank = self.trainer.global_rank
|
||||
|
||||
# set up server using proc 0's ip address
|
||||
# try to init for 20 times at max in case ports are taken
|
||||
# where to store ip_table
|
||||
model.trainer = self.trainer
|
||||
model.init_ddp_connection(
|
||||
self.trainer.global_rank,
|
||||
self.trainer.world_size,
|
||||
self.trainer.is_slurm_managing_tasks
|
||||
)
|
||||
|
||||
# call setup after the ddp process has connected
|
||||
self.trainer.call_setup_hook(model)
|
||||
|
||||
# on world_size=0 let everyone know training is starting
|
||||
if self.trainer.is_global_zero:
|
||||
log.info('-' * 100)
|
||||
log.info(f'distributed_backend={self.trainer.distributed_backend}')
|
||||
log.info(f'All DDP processes registered. Starting ddp with {self.trainer.world_size} processes')
|
||||
log.info('-' * 100)
|
||||
|
||||
# CHOOSE OPTIMIZER
|
||||
# allow for lr schedulers as well
|
||||
optimizers, lr_schedulers, optimizer_frequencies = self.trainer.init_optimizers(model)
|
||||
self.trainer.optimizers = optimizers
|
||||
self.trainer.lr_schedulers = lr_schedulers
|
||||
self.trainer.optimizer_frequencies = optimizer_frequencies
|
||||
|
||||
# MODEL
|
||||
# copy model to each gpu
|
||||
if self.trainer.on_gpu:
|
||||
gpu_idx = process_idx
|
||||
|
||||
# when using ddp, the master process (proc 0) continues running as the main one
|
||||
# this means that the local rank will always be 0
|
||||
# (even if cuda visible devices has other visible gpus)
|
||||
# this means that the master process needs to pull the 0th visible index as the device number
|
||||
if is_master:
|
||||
available_gpus = os.environ['CUDA_VISIBLE_DEVICES'].split(',')
|
||||
gpu_idx = int(available_gpus[self.trainer.local_rank])
|
||||
|
||||
self.trainer.root_gpu = gpu_idx
|
||||
torch.cuda.set_device(self.trainer.root_gpu)
|
||||
model.cuda(self.trainer.root_gpu)
|
||||
|
||||
# set model properties before going into wrapper
|
||||
self.trainer.copy_trainer_model_properties(model)
|
||||
|
||||
# AMP
|
||||
# run through amp wrapper before going to distributed DP
|
||||
# TODO: remove with dropping NVIDIA AMP support
|
||||
if self.trainer.use_amp and not NATIVE_AMP_AVALAIBLE:
|
||||
model, optimizers = model.configure_apex(amp, model, self.trainer.optimizers, self.trainer.amp_level)
|
||||
self.trainer.optimizers = optimizers
|
||||
self.trainer.reinit_scheduler_properties(self.trainer.optimizers, self.trainer.lr_schedulers)
|
||||
|
||||
# DDP2 uses all GPUs on the machine
|
||||
if self.trainer.distributed_backend == 'ddp' or self.trainer.distributed_backend == 'ddp_spawn':
|
||||
device_ids = [self.trainer.root_gpu]
|
||||
else: # includes ddp_cpu
|
||||
device_ids = None
|
||||
|
||||
# allow user to configure ddp
|
||||
model = model.configure_ddp(model, device_ids)
|
||||
|
||||
# continue training routine
|
||||
results = self.trainer.run_pretrain_routine(model)
|
||||
|
||||
# get original model
|
||||
model = self.trainer.get_model()
|
||||
|
||||
# persist info in ddp_spawn
|
||||
self.trainer.transfer_distrib_spawn_state_on_fit_end(model, mp_queue, results)
|
||||
|
||||
# clean up memory
|
||||
torch.cuda.empty_cache()
|
||||
|
||||
if self.trainer.global_rank == 0 and self.trainer.distributed_backend not in ['ddp_spawn', 'ddp_cpu']:
|
||||
return results
|
||||
|
|
@ -60,7 +60,7 @@ class DDPSpawnBackend(object):
|
|||
self.trainer.model = model
|
||||
return results
|
||||
|
||||
def ddp_train(self, process_idx, mp_queue, model, is_master=False, proc_offset=0):
|
||||
def ddp_train(self, process_idx, mp_queue, model):
|
||||
"""
|
||||
Entry point for ddp
|
||||
|
||||
|
|
@ -68,15 +68,10 @@ class DDPSpawnBackend(object):
|
|||
process_idx:
|
||||
mp_queue: multiprocessing queue
|
||||
model:
|
||||
is_master:
|
||||
proc_offset:
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
# offset the process id if requested
|
||||
process_idx = process_idx + proc_offset
|
||||
|
||||
# show progressbar only on progress_rank 0
|
||||
if (self.trainer.node_rank != 0 or process_idx != 0) and self.trainer.progress_bar_callback is not None:
|
||||
self.trainer.progress_bar_callback.disable()
|
||||
|
|
@ -126,11 +121,6 @@ class DDPSpawnBackend(object):
|
|||
# copy model to each gpu
|
||||
if self.trainer.on_gpu:
|
||||
gpu_idx = process_idx
|
||||
if is_master:
|
||||
# source of truth is cuda for gpu idx
|
||||
gpus = os.environ['CUDA_VISIBLE_DEVICES'].split(',')
|
||||
gpu_idx = int(gpus[self.trainer.local_rank])
|
||||
|
||||
self.trainer.root_gpu = gpu_idx
|
||||
torch.cuda.set_device(self.trainer.root_gpu)
|
||||
model.cuda(self.trainer.root_gpu)
|
||||
|
|
|
|||
|
|
@ -164,15 +164,6 @@ else:
|
|||
HOROVOD_AVAILABLE = True
|
||||
|
||||
|
||||
try:
|
||||
from hydra.utils import to_absolute_path, get_original_cwd
|
||||
from hydra.core.hydra_config import HydraConfig
|
||||
except ImportError:
|
||||
HYDRA_AVAILABLE = False
|
||||
else:
|
||||
HYDRA_AVAILABLE = True
|
||||
|
||||
|
||||
try:
|
||||
import torch_xla
|
||||
except ImportError:
|
||||
|
|
@ -205,6 +196,7 @@ class TrainerDDPMixin(ABC):
|
|||
node_rank: int
|
||||
tpu_cores: int
|
||||
testing: bool
|
||||
global_rank: int
|
||||
datamodule: Optional[LightningDataModule]
|
||||
|
||||
@property
|
||||
|
|
@ -429,175 +421,6 @@ class TrainerDDPMixin(ABC):
|
|||
|
||||
os.environ['MASTER_PORT'] = str(default_port)
|
||||
|
||||
def spawn_ddp_children(self, model):
|
||||
port = os.environ['MASTER_PORT']
|
||||
|
||||
master_address = '127.0.0.1' if 'MASTER_ADDR' not in os.environ else os.environ['MASTER_ADDR']
|
||||
os.environ['MASTER_PORT'] = f'{port}'
|
||||
os.environ['MASTER_ADDR'] = f'{master_address}'
|
||||
|
||||
# allow the user to pass the node rank
|
||||
node_rank = '0'
|
||||
if 'NODE_RANK' in os.environ:
|
||||
node_rank = os.environ['NODE_RANK']
|
||||
if 'GROUP_RANK' in os.environ:
|
||||
node_rank = os.environ['GROUP_RANK']
|
||||
|
||||
os.environ['NODE_RANK'] = node_rank
|
||||
os.environ['LOCAL_RANK'] = '0'
|
||||
|
||||
# when user is using hydra find the absolute path
|
||||
path_lib = abspath if not HYDRA_AVAILABLE else to_absolute_path
|
||||
|
||||
# pull out the commands used to run the script and resolve the abs file path
|
||||
command = sys.argv
|
||||
try:
|
||||
full_path = path_lib(command[0])
|
||||
except Exception as e:
|
||||
full_path = abspath(command[0])
|
||||
|
||||
command[0] = full_path
|
||||
# use the same python interpreter and actually running
|
||||
command = [sys.executable] + command
|
||||
|
||||
# since this script sets the visible devices we replace the gpus flag with a number
|
||||
num_gpus = os.environ['CUDA_VISIBLE_DEVICES'].split(',').__len__()
|
||||
|
||||
if '--gpus' in command:
|
||||
gpu_flag_idx = command.index('--gpus')
|
||||
command[gpu_flag_idx + 1] = f'{num_gpus}'
|
||||
|
||||
os.environ['WORLD_SIZE'] = f'{num_gpus * self.num_nodes}'
|
||||
|
||||
self.interactive_ddp_procs = []
|
||||
for local_rank in range(1, self.num_processes):
|
||||
env_copy = os.environ.copy()
|
||||
env_copy['LOCAL_RANK'] = f'{local_rank}'
|
||||
|
||||
# start process
|
||||
# if hydra is available and initialized, make sure to set the cwd correctly
|
||||
cwd: Optional[str] = None
|
||||
if HYDRA_AVAILABLE:
|
||||
if HydraConfig.initialized():
|
||||
cwd = get_original_cwd()
|
||||
proc = subprocess.Popen(command, env=env_copy, cwd=cwd)
|
||||
self.interactive_ddp_procs.append(proc)
|
||||
|
||||
# starting all processes at once can cause issues
|
||||
# with dataloaders delay between 1-10 seconds
|
||||
delay = np.random.uniform(1, 5, 1)[0]
|
||||
sleep(delay)
|
||||
|
||||
local_rank = 0
|
||||
results = self.ddp_train(local_rank, mp_queue=None, model=model, is_master=True)
|
||||
del os.environ['WORLD_SIZE']
|
||||
|
||||
return results
|
||||
|
||||
def ddp_train(self, process_idx, mp_queue, model, is_master=False, proc_offset=0):
|
||||
"""
|
||||
Entry point for ddp
|
||||
|
||||
Args:
|
||||
process_idx:
|
||||
mp_queue: multiprocessing queue
|
||||
model:
|
||||
is_master:
|
||||
proc_offset:
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
# offset the process id if requested
|
||||
process_idx = process_idx + proc_offset
|
||||
|
||||
# show progressbar only on progress_rank 0
|
||||
if (self.node_rank != 0 or process_idx != 0) and self.progress_bar_callback is not None:
|
||||
self.progress_bar_callback.disable()
|
||||
|
||||
# determine which process we are and world size
|
||||
if self.use_ddp:
|
||||
self.local_rank = process_idx
|
||||
self.global_rank = self.node_rank * self.num_processes + process_idx
|
||||
self.world_size = self.num_nodes * self.num_processes
|
||||
|
||||
elif self.use_ddp2:
|
||||
self.local_rank = self.node_rank
|
||||
self.global_rank = self.node_rank
|
||||
self.world_size = self.num_nodes
|
||||
|
||||
# set warning rank
|
||||
rank_zero_only.rank = self.global_rank
|
||||
|
||||
# set up server using proc 0's ip address
|
||||
# try to init for 20 times at max in case ports are taken
|
||||
# where to store ip_table
|
||||
model.trainer = self
|
||||
model.init_ddp_connection(self.global_rank, self.world_size, self.is_slurm_managing_tasks)
|
||||
|
||||
# call setup after the ddp process has connected
|
||||
self.call_setup_hook(model)
|
||||
|
||||
# on world_size=0 let everyone know training is starting
|
||||
if self.is_global_zero:
|
||||
log.info('-' * 100)
|
||||
log.info(f'distributed_backend={self.distributed_backend}')
|
||||
log.info(f'All DDP processes registered. Starting ddp with {self.world_size} processes')
|
||||
log.info('-' * 100)
|
||||
|
||||
# CHOOSE OPTIMIZER
|
||||
# allow for lr schedulers as well
|
||||
self.optimizers, self.lr_schedulers, self.optimizer_frequencies = self.init_optimizers(model)
|
||||
|
||||
# MODEL
|
||||
# copy model to each gpu
|
||||
if self.on_gpu:
|
||||
gpu_idx = process_idx
|
||||
if is_master:
|
||||
# source of truth is cuda for gpu idx
|
||||
gpu_idx = self.local_rank
|
||||
|
||||
self.root_gpu = gpu_idx
|
||||
torch.cuda.set_device(self.root_gpu)
|
||||
model.cuda(self.root_gpu)
|
||||
|
||||
# set model properties before going into wrapper
|
||||
self.copy_trainer_model_properties(model)
|
||||
|
||||
# AMP
|
||||
# run through amp wrapper before going to distributed DP
|
||||
# TODO: remove with dropping NVIDIA AMP support
|
||||
if self.use_amp and not NATIVE_AMP_AVALAIBLE:
|
||||
model, optimizers = model.configure_apex(amp, model, self.optimizers, self.amp_level)
|
||||
self.optimizers = optimizers
|
||||
self.reinit_scheduler_properties(self.optimizers, self.lr_schedulers)
|
||||
|
||||
# DDP2 uses all GPUs on the machine
|
||||
if self.distributed_backend == 'ddp' or self.distributed_backend == 'ddp_spawn':
|
||||
device_ids = [self.root_gpu]
|
||||
elif self.use_ddp2:
|
||||
device_ids = self.data_parallel_device_ids
|
||||
else: # includes ddp_cpu
|
||||
device_ids = None
|
||||
|
||||
# allow user to configure ddp
|
||||
model = model.configure_ddp(model, device_ids)
|
||||
|
||||
# continue training routine
|
||||
results = self.run_pretrain_routine(model)
|
||||
|
||||
# get original model
|
||||
model = self.get_model()
|
||||
|
||||
# persist info in ddp_spawn
|
||||
self.transfer_distrib_spawn_state_on_fit_end(model, mp_queue, results)
|
||||
|
||||
# clean up memory
|
||||
torch.cuda.empty_cache()
|
||||
|
||||
if self.global_rank == 0 and self.distributed_backend not in ['ddp_spawn', 'ddp_cpu']:
|
||||
return results
|
||||
|
||||
def transfer_distrib_spawn_state_on_fit_end(self, model, mp_queue, results):
|
||||
if self.distributed_backend.lower() not in ['ddp_spawn', 'ddp_cpu', 'tpu']:
|
||||
return
|
||||
|
|
|
|||
|
|
@ -52,7 +52,7 @@ from pytorch_lightning.utilities.debugging import InternalDebugger
|
|||
from pytorch_lightning.utilities.exceptions import MisconfigurationException
|
||||
from pytorch_lightning.trainer.configuration_validator import ConfigValidator
|
||||
from pytorch_lightning.accelerator_backends import (
|
||||
GPUBackend, TPUBackend, CPUBackend, DDPSpawnBackend, DataParallelBackend)
|
||||
GPUBackend, TPUBackend, CPUBackend, DDPSpawnBackend, DataParallelBackend, DDPBackend, DDP2Backend)
|
||||
|
||||
# warnings to ignore in trainer
|
||||
warnings.filterwarnings(
|
||||
|
|
@ -972,48 +972,54 @@ class Trainer(
|
|||
self._run_lr_finder_internally(model)
|
||||
model.logger = self.logger # reset logger binding
|
||||
|
||||
# route to appropriate start method
|
||||
# when using multi-node or DDP within a node start each module in a separate process
|
||||
# set testing if set in environ
|
||||
self.testing = os.environ.get('PL_TESTING_MODE', self.testing)
|
||||
|
||||
# -------------------
|
||||
# determine ddp mode
|
||||
# -------------------
|
||||
# SLURM ddp
|
||||
use_slurm_ddp = self.use_ddp and self.is_slurm_managing_tasks
|
||||
|
||||
# torchelastic or general non_slurm ddp
|
||||
te_flags_passed = 'WORLD_SIZE' in os.environ and ('GROUP_RANK' in os.environ or 'NODE_RANK' in os.environ)
|
||||
use_torchelastic_ddp = self.use_ddp and te_flags_passed
|
||||
|
||||
use_ddp_spawn = self.use_ddp and self.distributed_backend in ['ddp_cpu', 'ddp_spawn']
|
||||
|
||||
# -------------------
|
||||
# route training mode
|
||||
# -------------------
|
||||
# DDP2 (cluster only)
|
||||
if self.use_ddp2:
|
||||
if self.is_slurm_managing_tasks:
|
||||
task = int(os.environ['SLURM_LOCALID'])
|
||||
self.accelerator_backend = DDP2Backend(self)
|
||||
self.accelerator_backend.setup()
|
||||
self.accelerator_backend.train(model)
|
||||
|
||||
# torchelastic or general non_slurm ddp2
|
||||
elif 'WORLD_SIZE' in os.environ and ('GROUP_RANK' in os.environ or 'NODE_RANK' in os.environ):
|
||||
task = int(os.environ['LOCAL_RANK'])
|
||||
elif use_slurm_ddp:
|
||||
self.accelerator_backend = DDPBackend(self)
|
||||
self.accelerator_backend.slurm_setup()
|
||||
self.accelerator_backend.train(model)
|
||||
|
||||
self.ddp_train(process_idx=task, mp_queue=None, model=model)
|
||||
elif use_torchelastic_ddp:
|
||||
self.accelerator_backend = DDPBackend(self)
|
||||
self.accelerator_backend.torchelastic_setup()
|
||||
self.accelerator_backend.train(model)
|
||||
|
||||
elif self.use_ddp:
|
||||
# regular ddp using .spawn
|
||||
elif use_ddp_spawn:
|
||||
self.accelerator_backend = DDPSpawnBackend(self)
|
||||
self.accelerator_backend.setup()
|
||||
self.accelerator_backend.train(model, nprocs=self.num_processes)
|
||||
results = self.accelerator_backend.teardown(model)
|
||||
|
||||
# set testing if set in environ
|
||||
self.testing = os.environ.get('PL_TESTING_MODE', self.testing)
|
||||
|
||||
if self.is_slurm_managing_tasks:
|
||||
task = int(os.environ['SLURM_LOCALID'])
|
||||
self.ddp_train(process_idx=task, mp_queue=None, model=model)
|
||||
|
||||
# torchelastic or general non_slurm ddp
|
||||
elif 'WORLD_SIZE' in os.environ and ('GROUP_RANK' in os.environ or 'NODE_RANK' in os.environ):
|
||||
task = int(os.environ['LOCAL_RANK'])
|
||||
self.ddp_train(process_idx=task, mp_queue=None, model=model)
|
||||
|
||||
elif self.distributed_backend == 'ddp_cpu':
|
||||
self.accelerator_backend = DDPSpawnBackend(self)
|
||||
self.accelerator_backend.setup()
|
||||
self.accelerator_backend.train(model, nprocs=self.num_processes)
|
||||
results = self.accelerator_backend.teardown(model)
|
||||
|
||||
elif self.distributed_backend == 'ddp_spawn':
|
||||
self.accelerator_backend = DDPSpawnBackend(self)
|
||||
self.accelerator_backend.setup()
|
||||
self.accelerator_backend.train(model, nprocs=self.num_processes)
|
||||
results = self.accelerator_backend.teardown(model)
|
||||
|
||||
elif self.distributed_backend == 'ddp':
|
||||
self.set_random_port()
|
||||
results = self.spawn_ddp_children(model)
|
||||
# ddp
|
||||
elif self.distributed_backend == 'ddp':
|
||||
self.set_random_port()
|
||||
self.accelerator_backend = DDPBackend(self)
|
||||
results = self.accelerator_backend.spawn_ddp_children(model)
|
||||
|
||||
# dp
|
||||
elif self.use_dp:
|
||||
self.accelerator_backend = DataParallelBackend(self)
|
||||
self.accelerator_backend.setup(model)
|
||||
|
|
|
|||
Loading…
Reference in New Issue