You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
ColossalAI/colossalai/auto_parallel/passes/runtime_preparation_pass.py

224 lines
11 KiB

from copy import deepcopy
from typing import List
import torch
from torch.fx import symbolic_trace
from torch.fx.node import Node
from colossalai.auto_parallel.tensor_shard.sharding_strategy import (
CommAction,
CommType,
OperationDataType,
ShardingStrategy,
)
from colossalai.device.device_mesh import DeviceMesh
from colossalai.tensor.comm_spec import _all_reduce
from colossalai.tensor.shape_consistency import ShapeConsistencyManager
from colossalai.tensor.sharding_spec import ShardingSpec
shape_consistency_manager = ShapeConsistencyManager()
def _solution_annotatation(gm: torch.fx.GraphModule, solution: List[int]):
"""
This method is used to stick the solution strategy to the nodes and add the information
required in runtime into graph as placeholder nodes.
"""
mod_graph = gm.graph
nodes = tuple(mod_graph.nodes)
# the dict to get origin sharding spec of node
origin_node_sharding_spec_dict = {}
for node_index, (node, strategy_index) in enumerate(zip(nodes, solution)):
strategies_vector = node.strategies_vector
# stick the solution strategy to the corresponding node
setattr(node, 'best_strategy', strategies_vector[strategy_index])
setattr(node, 'sharding_spec', strategies_vector[strategy_index].get_sharding_spec_by_name(str(node)))
origin_node_sharding_spec_dict[node_index] = strategies_vector[strategy_index].get_sharding_spec_by_name(
str(node))
# experimental pass for torch.Tensor.view
# Arguments of view op will be divided in the sharded dimensions.
for node in nodes:
if node.op == 'call_method' and getattr(node.args[0]._meta_data.__class__, node.target) in (torch.Tensor.view,):
output_dim_partition_dict = node.sharding_spec.dim_partition_dict
device_mesh = node.sharding_spec.device_mesh
new_args = []
for arg in node.args:
if isinstance(arg, Node):
if isinstance(arg._meta_data, int):
new_args.append(arg._meta_data)
else:
new_args.append(arg)
else:
assert isinstance(arg, int), 'The argument in view node should be either type of Node or int.'
new_args.append(arg)
for dim, shard_dims in output_dim_partition_dict.items():
total_shard_size = 1
for shard_dim in shard_dims:
total_shard_size *= device_mesh.shape[shard_dim]
new_args[dim + 1] //= total_shard_size
node.args = tuple(new_args)
# the dict to get input sharding specs of user node
sharding_spec_convert_dict = {}
# the dict to record comm actions of nodes
comm_actions_dict = {}
for index, node in enumerate(nodes):
target_sharding_specs = []
for user_node in node.strategies_vector.successor_nodes:
target_sharding_spec = user_node.best_strategy.get_sharding_spec_by_name(str(node.name))
target_sharding_specs.append(target_sharding_spec)
sharding_spec_convert_dict[index] = target_sharding_specs
setattr(node, 'target_sharding_specs', target_sharding_specs)
# the get_attr node strategy is kind of pending strategy, which means we will change it
# to the same strategy of the user node.
if node.op == 'get_attr':
assert len(target_sharding_specs) == 1, f'sharing weight is not supported in current version.'
new_sharding_spec = target_sharding_specs[0]
user_strategy = node.strategies_vector.successor_nodes[0].best_strategy
op_data_in_user = user_strategy.get_op_data_by_name(str(node))
origin_node_sharding_spec_dict[index] = new_sharding_spec
origin_pending_strategy = node.best_strategy
origin_op_data = origin_pending_strategy.get_op_data_by_name(str(node))
new_sharding_specs = origin_pending_strategy.sharding_specs
new_sharding_specs[origin_op_data] = new_sharding_spec
new_communication_actions = {}
if op_data_in_user in user_strategy.communication_actions:
new_communication_action = user_strategy.communication_actions.pop(op_data_in_user)
new_communication_action.arg_index = 0
new_communication_actions[origin_op_data] = new_communication_action
new_strategy = ShardingStrategy(name=str(new_sharding_spec.sharding_sequence),
sharding_specs=new_sharding_specs,
compute_cost=origin_pending_strategy.compute_cost,
communication_cost=origin_pending_strategy.communication_cost,
memory_cost=origin_pending_strategy.memory_cost,
communication_actions=new_communication_actions)
setattr(node, 'best_strategy', new_strategy)
setattr(node, 'sharding_spec', new_sharding_spec)
comm_action_dict = {}
for op_data, comm_action in node.best_strategy.communication_actions.items():
comm_action_dict[op_data.name] = comm_action
comm_actions_dict[index] = comm_action_dict
# add above dicts into graph
for node in nodes:
if node.op != 'placeholder':
with mod_graph.inserting_before(node):
input_specs_node = mod_graph.create_node('placeholder', target='sharding_spec_convert_dict')
origin_specs_node = mod_graph.create_node('placeholder', target='origin_node_sharding_spec_dict')
comm_actions_dict_node = mod_graph.create_node('placeholder', target='comm_actions_dict')
break
return gm, sharding_spec_convert_dict, origin_node_sharding_spec_dict, comm_actions_dict
def _module_params_sharding(gm: torch.fx.GraphModule, device_mesh):
"""
Apply the sharding action to the module parameters and buffers following the
instructions of solver solution.
"""
mod_graph = gm.graph
nodes = tuple(mod_graph.nodes)
# This stream is created for overlaping the communication and computation.
reduction_stream = torch.cuda.Stream()
for node in nodes:
if node.op == 'call_module':
target_module = node.graph.owning_module.get_submodule(node.target)
for name, param in target_module.named_parameters():
target_sharding_spec = node.best_strategy.get_sharding_spec_by_name(name)
# apply the sharding spec of parameters
if target_sharding_spec.dim_partition_dict != {}:
origin_sharding_spec = ShardingSpec(device_mesh, param.shape, {})
setattr(param, 'sharding_spec', origin_sharding_spec)
# TODO: build a ColoParamter class to manager the distributed parameters
param_sharded = torch.nn.Parameter(
shape_consistency_manager.apply_for_autoparallel_runtime(param.data, param.sharding_spec,
target_sharding_spec).detach().clone())
else:
param_sharded = param
setattr(target_module, name, param_sharded)
comm_actions = node.best_strategy.communication_actions
for operation_data, comm_action in comm_actions.items():
comm_spec_to_use = comm_action.comm_spec
# register hook to the parameters
if operation_data.type == OperationDataType.PARAM and operation_data.name == name and comm_action.comm_type == CommType.HOOK:
def wrapper(param, comm_spec):
def hook_fn(grad):
_all_reduce(grad, comm_spec, async_op=False)
param.register_hook(hook_fn)
wrapper(param_sharded, comm_spec_to_use)
sharded_buffer_dict = {}
# apply the sharding spec of buffers
for name, buffer in target_module.named_buffers():
origin_sharding_spec = ShardingSpec(device_mesh, buffer.shape, {})
setattr(buffer, 'sharding_spec', origin_sharding_spec)
target_sharding_spec = node.best_strategy.get_sharding_spec_by_name(name)
buffer_sharded = shape_consistency_manager.apply(buffer, target_sharding_spec)
sharded_buffer_dict[name] = buffer_sharded
for name, buffer_sharded in sharded_buffer_dict.items():
setattr(target_module, name, buffer_sharded.detach().clone())
if node.op == 'get_attr':
root = node.graph.owning_module
atoms = node.target.split(".")
attr_len = len(atoms)
if attr_len == 1:
target_module = root
target = getattr(root, atoms[0])
else:
target_module = root.get_submodule(atoms[-2])
target = getattr(target_module, atoms[-1])
target_sharding_spec = node.sharding_spec
if target_sharding_spec.dim_partition_dict != {}:
origin_sharding_spec = ShardingSpec(device_mesh, target.shape, {})
setattr(target, 'sharding_spec', origin_sharding_spec)
# TODO: build a ColoParamter class to manager the distributed parameters
target_sharded = torch.nn.Parameter(
shape_consistency_manager.apply_for_autoparallel_runtime(target.data, target.sharding_spec,
target_sharding_spec).detach().clone())
else:
target_sharded = target
setattr(target_module, atoms[-1], target_sharded)
comm_actions = node.best_strategy.communication_actions
for operation_data, comm_action in comm_actions.items():
comm_spec_to_use = comm_action.comm_spec
# register hook to the parameters
if isinstance(node._meta_data, torch.nn.parameter.Parameter) and comm_action.comm_type == CommType.HOOK:
def wrapper(param, comm_spec):
def hook_fn(grad):
_all_reduce(grad, comm_spec, async_op=False)
param.register_hook(hook_fn)
wrapper(target_sharded, comm_spec_to_use)
return gm
def implicit_comm_action_apply(gm: torch.fx.GraphModule):
"""
replace the origin kernel into kernel with implicit communication inside.
"""
pass
def runtime_preparation_pass(gm: torch.fx.GraphModule, solution: List[int], device_mesh: DeviceMesh):
gm, sharding_spec_convert_dict, origin_node_sharding_spec_dict, comm_actions_dict = _solution_annotatation(
gm, solution)
# TODO: the pass below should be uncommented after the implementation of implicit_comm_action_apply_pass completed.
# gm = implicit_comm_action_apply(gm)
gm = _module_params_sharding(gm, device_mesh)
return gm, sharding_spec_convert_dict, origin_node_sharding_spec_dict, comm_actions_dict