2022-08-30 08:32:09 +00:00
|
|
|
from torch.fx import Graph, Node
|
|
|
|
|
from colossalai.tensor.sharding_spec import ShardingSpec
|
2022-09-13 06:47:09 +00:00
|
|
|
from colossalai.device.device_mesh import DeviceMesh
|
|
|
|
|
from colossalai.tensor.shape_consistency import ShapeConsistencyManager
|
|
|
|
|
from .options import SolverOptions
|
2022-09-13 04:07:09 +00:00
|
|
|
from . import ShardingStrategy, StrategiesVector
|
|
|
|
|
from .op_handler import *
|
2022-08-30 08:32:09 +00:00
|
|
|
from .constants import *
|
|
|
|
|
from copy import deepcopy
|
|
|
|
|
import math
|
|
|
|
|
import torch
|
|
|
|
|
import operator
|
|
|
|
|
from typing import Dict, List
|
2022-09-13 10:30:18 +00:00
|
|
|
from ._utils import generate_sharding_spec, generate_resharding_costs
|
2022-08-30 08:32:09 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
class StrategiesConstructor:
|
2022-09-13 06:47:09 +00:00
|
|
|
"""
|
|
|
|
|
StrategiesConstructor is used to construct the parallelization plan for the model execution.
|
|
|
|
|
|
|
|
|
|
Args:
|
|
|
|
|
graph (Graph): a Graph object used for analysis and strategy generation.
|
|
|
|
|
device_mesh (DeviceMesh): a DeviceMesh object which contains the meta information about the cluster.
|
|
|
|
|
solver_options (SolverOptions): a SolverOptions object which specifies the preferences for plan searching.
|
|
|
|
|
"""
|
|
|
|
|
|
2022-09-13 10:30:18 +00:00
|
|
|
def __init__(self, graph: Graph, device_mesh: DeviceMesh, solver_options: SolverOptions):
|
2022-08-30 08:32:09 +00:00
|
|
|
self.graph = graph
|
2022-09-13 06:47:09 +00:00
|
|
|
assert graph.owning_module is not None, 'The given graph is not associated with a owning_module'
|
2022-08-30 08:32:09 +00:00
|
|
|
self.root_module = self.graph.owning_module
|
|
|
|
|
self.nodes = list(graph.nodes)
|
|
|
|
|
self.device_mesh = device_mesh
|
|
|
|
|
self.leaf_strategies = []
|
|
|
|
|
self.strategy_map = {}
|
|
|
|
|
self.solver_options = solver_options
|
|
|
|
|
|
|
|
|
|
def remove_duplicated_strategy(self, strategies_vector):
|
|
|
|
|
'''
|
|
|
|
|
In build_strategies_and_cost method, we may produce some duplicated strategies.
|
|
|
|
|
In this method, we will remove the duplicated strategies depending on the strategies name.
|
|
|
|
|
'''
|
|
|
|
|
name_checklist = []
|
|
|
|
|
remove_list = []
|
|
|
|
|
for strategy in strategies_vector:
|
|
|
|
|
if strategy.name not in name_checklist:
|
|
|
|
|
name_checklist.append(strategy.name)
|
|
|
|
|
else:
|
|
|
|
|
remove_list.append(strategy)
|
|
|
|
|
for strategy in remove_list:
|
|
|
|
|
strategies_vector.remove(strategy)
|
|
|
|
|
|
|
|
|
|
def build_strategies_and_cost(self):
|
|
|
|
|
for node in self.nodes:
|
|
|
|
|
strategies_vector = StrategiesVector(node)
|
|
|
|
|
# placeholder node
|
|
|
|
|
if node.op == 'placeholder':
|
2022-09-13 06:47:09 +00:00
|
|
|
# For placeholder nodes, if solver_options.fast is True, we just let them in
|
2022-08-30 08:32:09 +00:00
|
|
|
# fully replicate status, then strategies of following node will be treated equally due
|
|
|
|
|
# to replicate status has no resharding cost to other status. At the same time, the searching
|
|
|
|
|
# space is smaller than enumerating all the possible sharding spec for the placeholder node.
|
|
|
|
|
# Otherwise, all the possible sharding spec for the placeholder node will be enumerated.
|
|
|
|
|
|
2022-09-13 06:47:09 +00:00
|
|
|
if self.solver_options.fast:
|
2022-08-30 08:32:09 +00:00
|
|
|
# create sharding strategy for placeholder
|
|
|
|
|
name = 'Replica Placeholder'
|
|
|
|
|
dim_partition_dict = {}
|
2022-09-13 07:43:22 +00:00
|
|
|
output_sharding_spec = generate_sharding_spec(node, self.device_mesh, dim_partition_dict)
|
2022-08-30 08:32:09 +00:00
|
|
|
# TODO: use meta_info_prop to profile memory cost
|
|
|
|
|
memory_cost = 0
|
|
|
|
|
sharding_strategy_placeholder = ShardingStrategy(name,
|
|
|
|
|
output_sharding_spec,
|
|
|
|
|
memory_cost=memory_cost)
|
|
|
|
|
strategies_vector.append(sharding_strategy_placeholder)
|
|
|
|
|
|
|
|
|
|
# get_attr node
|
|
|
|
|
if node.op == 'get_attr':
|
2022-09-13 06:47:09 +00:00
|
|
|
# Same as placeholder nodes, if solver_options.fast is True, we just let them in
|
2022-08-30 08:32:09 +00:00
|
|
|
# fully replicate status, then strategies of following node will be treated equally due
|
|
|
|
|
# to replicate status has no resharding cost to other status. At the same time, the searching
|
|
|
|
|
# space is smaller than enumerating all the possible sharding spec for the get_attr node.
|
|
|
|
|
# Otherwise, all the possible sharding spec for the get_attr node will be enumerated.
|
2022-09-13 06:47:09 +00:00
|
|
|
if self.solver_options.fast:
|
2022-08-30 08:32:09 +00:00
|
|
|
# create sharding strategy for get_attr
|
|
|
|
|
name = 'Replica Attribute'
|
|
|
|
|
dim_partition_dict = {}
|
2022-09-13 07:43:22 +00:00
|
|
|
output_sharding_spec = generate_sharding_spec(node, self.device_mesh, dim_partition_dict)
|
2022-08-30 08:32:09 +00:00
|
|
|
# TODO: use meta_info_prop to profile memory cost
|
|
|
|
|
memory_cost = 0
|
|
|
|
|
sharding_strategy_attribute = ShardingStrategy(name, output_sharding_spec, memory_cost=memory_cost)
|
|
|
|
|
strategies_vector.append(sharding_strategy_attribute)
|
|
|
|
|
|
|
|
|
|
# call_module node
|
|
|
|
|
if node.op == 'call_module':
|
|
|
|
|
|
|
|
|
|
target = node.target
|
|
|
|
|
submod = self.root_module.get_submodule(target)
|
|
|
|
|
submod_type = type(submod)
|
|
|
|
|
|
|
|
|
|
# conv module
|
|
|
|
|
if submod_type in CONV_MODULE_OP:
|
|
|
|
|
# use ConvHandler to create sharding strategies for conv module node
|
2022-09-13 10:30:18 +00:00
|
|
|
conv_handler = ConvHandler(node, self.device_mesh, strategies_vector)
|
2022-08-30 08:32:09 +00:00
|
|
|
conv_handler.register_strategy()
|
|
|
|
|
|
|
|
|
|
# linear module
|
|
|
|
|
elif submod_type in LINEAR_MODULE_OP:
|
|
|
|
|
# use DotHandler to create sharding strategies for linear module node
|
2022-09-13 10:30:18 +00:00
|
|
|
dot_handler = DotHandler(node, self.device_mesh, strategies_vector)
|
2022-08-30 08:32:09 +00:00
|
|
|
dot_handler.register_strategy()
|
|
|
|
|
|
|
|
|
|
# element-wise module
|
|
|
|
|
elif submod_type in ELEMENTWISE_MODULE_OP:
|
|
|
|
|
# create sharding strategy for element-wise module
|
|
|
|
|
assert len(strategies_vector.predecessor_nodes
|
|
|
|
|
) == 1, f'Temporally, we just support single input element-wise op.'
|
|
|
|
|
input_node = strategies_vector.predecessor_nodes[0]
|
|
|
|
|
# For element-wise module, we keep the sharding spec of output node same as
|
|
|
|
|
# the input. Therefore, the different strategies of input node with same
|
|
|
|
|
# output sharding spec will generate same strategy for element-wise module.
|
|
|
|
|
sharding_spec_checklist = []
|
|
|
|
|
for strategy in input_node.strategies_vector:
|
|
|
|
|
# It looks a little bit confusing, the input of the processing node
|
|
|
|
|
# is the output of the input_node.
|
|
|
|
|
input_sharding_spec = strategy.output_sharding_spec
|
|
|
|
|
assert isinstance(input_sharding_spec,
|
|
|
|
|
ShardingSpec), f'The input node should NOT be a tuple of tensor.'
|
|
|
|
|
if input_sharding_spec in sharding_spec_checklist:
|
|
|
|
|
continue
|
|
|
|
|
|
|
|
|
|
sharding_spec_checklist.append(input_sharding_spec)
|
|
|
|
|
dim_partition_dict = deepcopy(input_sharding_spec.dim_partition_dict)
|
2022-09-13 07:43:22 +00:00
|
|
|
output_sharding_spec = generate_sharding_spec(node, self.device_mesh, dim_partition_dict)
|
2022-08-30 08:32:09 +00:00
|
|
|
|
|
|
|
|
name = f'{input_sharding_spec.sharding_sequence} -> {output_sharding_spec.sharding_sequence}'
|
|
|
|
|
|
|
|
|
|
# TODO: use meta_info_prop to profile memory cost and compute cost
|
|
|
|
|
compute_cost = node._meta_data.numel()
|
|
|
|
|
memory_cost = 0
|
2022-09-13 10:30:18 +00:00
|
|
|
resharding_costs = generate_resharding_costs(strategies_vector.predecessor_nodes,
|
|
|
|
|
[input_sharding_spec])
|
2022-08-30 08:32:09 +00:00
|
|
|
|
|
|
|
|
# to prevent the resharding happening, set their resharding cost to inf.
|
|
|
|
|
resharding_costs[input_node] = [
|
|
|
|
|
cost if cost == 0 else math.inf for cost in resharding_costs[input_node]
|
|
|
|
|
]
|
|
|
|
|
sharding_strategy = ShardingStrategy(name,
|
|
|
|
|
output_sharding_spec,
|
|
|
|
|
compute_cost=compute_cost,
|
|
|
|
|
memory_cost=memory_cost,
|
|
|
|
|
resharding_costs=resharding_costs,
|
|
|
|
|
input_shardings=[input_sharding_spec])
|
|
|
|
|
strategies_vector.append(sharding_strategy)
|
|
|
|
|
|
2022-09-13 04:07:09 +00:00
|
|
|
# BatchNormNd module
|
|
|
|
|
elif submod_type in BATCHNORM_MODULE_OP:
|
|
|
|
|
# bn1 call_module bn1 (conv1,)
|
|
|
|
|
# print(node, node.op, node.target, node.args)
|
|
|
|
|
# create sharding strategy for element-wise module
|
|
|
|
|
# input_node = strategies_vector.predecessor_nodes[0]
|
2022-09-14 02:25:45 +00:00
|
|
|
norm_handler = BatchNormHandler(node, self.device_mesh, strategies_vector)
|
2022-09-13 04:07:09 +00:00
|
|
|
norm_handler.register_strategy()
|
|
|
|
|
# for strategy in norm_handler.strategies_vector:
|
|
|
|
|
# print(f'{strategy.name}, computation_cost: {strategy.compute_cost}, memory_cost: {strategy.memory_cost}')
|
|
|
|
|
# assert False
|
|
|
|
|
|
|
|
|
|
# MaxPool module
|
|
|
|
|
elif submod_type in POOL_MODULE_OP:
|
|
|
|
|
# create sharding strategy for element-wise module
|
|
|
|
|
assert len(strategies_vector.predecessor_nodes
|
|
|
|
|
) == 1, f'Temporally, we just support single input element-wise op.'
|
|
|
|
|
input_node = strategies_vector.predecessor_nodes[0]
|
|
|
|
|
# For element-wise module, we keep the sharding spec of output node same as
|
|
|
|
|
# the input. Therefore, the different strategies of input node with same
|
|
|
|
|
# output sharding spec will generate same strategy for element-wise module.
|
|
|
|
|
sharding_spec_checklist = []
|
|
|
|
|
for strategy in input_node.strategies_vector:
|
|
|
|
|
# It looks a little bit confusing, the input of the processing node
|
|
|
|
|
# is the output of the input_node.
|
|
|
|
|
input_sharding_spec = strategy.output_sharding_spec
|
|
|
|
|
assert isinstance(input_sharding_spec,
|
|
|
|
|
ShardingSpec), f'The input node should NOT be a tuple of tensor.'
|
|
|
|
|
if input_sharding_spec in sharding_spec_checklist:
|
|
|
|
|
continue
|
|
|
|
|
|
|
|
|
|
sharding_spec_checklist.append(input_sharding_spec)
|
|
|
|
|
dim_partition_dict = deepcopy(input_sharding_spec.dim_partition_dict)
|
2022-09-13 07:43:22 +00:00
|
|
|
output_sharding_spec = generate_sharding_spec(node, self.device_mesh, dim_partition_dict)
|
2022-09-13 04:07:09 +00:00
|
|
|
|
|
|
|
|
name = f'{input_sharding_spec.sharding_sequence} -> {output_sharding_spec.sharding_sequence}'
|
|
|
|
|
|
|
|
|
|
# TODO: use meta_info_prop to profile memory cost and compute cost
|
|
|
|
|
compute_cost = node._meta_data.numel()
|
|
|
|
|
memory_cost = 0
|
2022-09-13 10:30:18 +00:00
|
|
|
resharding_costs = generate_resharding_costs(strategies_vector.predecessor_nodes,
|
|
|
|
|
[input_sharding_spec])
|
2022-09-13 04:07:09 +00:00
|
|
|
|
|
|
|
|
sharding_strategy = ShardingStrategy(name,
|
|
|
|
|
output_sharding_spec,
|
|
|
|
|
compute_cost=compute_cost,
|
|
|
|
|
memory_cost=memory_cost,
|
|
|
|
|
resharding_costs=resharding_costs,
|
|
|
|
|
input_shardings=[input_sharding_spec])
|
|
|
|
|
strategies_vector.append(sharding_strategy)
|
|
|
|
|
|
2022-08-30 08:32:09 +00:00
|
|
|
# other module
|
|
|
|
|
else:
|
|
|
|
|
raise RuntimeError(f'{submod_type} module is NOT supported now.')
|
|
|
|
|
|
|
|
|
|
# call_function node
|
|
|
|
|
if node.op == 'call_function':
|
|
|
|
|
target = node.target
|
|
|
|
|
# conv function
|
|
|
|
|
if target in CONV_FUNC_OP:
|
|
|
|
|
# use ConvHandler to create sharding strategies for conv node
|
|
|
|
|
# TODO: the operator_handler does NOT support function node processing now.
|
2022-09-14 02:25:45 +00:00
|
|
|
conv_handler = ConvHandler(node, self.device_mesh, strategies_vector)
|
2022-08-30 08:32:09 +00:00
|
|
|
conv_handler.register_strategy()
|
|
|
|
|
|
|
|
|
|
# linear function
|
|
|
|
|
elif target in LINEAR_FUNC_OP:
|
|
|
|
|
# use DotHandler to create sharding strategies for linear node
|
|
|
|
|
# TODO: the operator_handler does NOT support function node processing now.
|
2022-09-14 02:25:45 +00:00
|
|
|
linear_handler = DotHandler(node, self.device_mesh, strategies_vector)
|
2022-08-30 08:32:09 +00:00
|
|
|
linear_handler.register_strategy()
|
|
|
|
|
|
2022-09-14 02:25:45 +00:00
|
|
|
# reshape function
|
|
|
|
|
elif target in RESHAPE_FUNC_OP:
|
|
|
|
|
# use ReshapeHandler to create sharding strategies for rehsape node
|
|
|
|
|
reshape_handler = ReshapeHandler(node, self.device_mesh, strategies_vector)
|
|
|
|
|
reshape_handler.register_strategy()
|
|
|
|
|
|
2022-08-30 08:32:09 +00:00
|
|
|
# element-wise function
|
|
|
|
|
elif target in ELEMENTWISE_FUNC_OP:
|
|
|
|
|
# TODO: integrate element-wise func and module together
|
|
|
|
|
# create sharding strategy for element-wise function
|
|
|
|
|
assert len(strategies_vector.predecessor_nodes
|
2022-09-13 04:07:09 +00:00
|
|
|
) == 1, f'Temporally, we just support single input element-wise op, node name is {node}.'
|
2022-08-30 08:32:09 +00:00
|
|
|
input_node = strategies_vector.predecessor_nodes[0]
|
|
|
|
|
# For element-wise function, we keep the sharding spec of output node same as
|
|
|
|
|
# the input. Therefore, the different strategies of input node with same
|
|
|
|
|
# output sharding spec will generate same strategy for element-wise function.
|
|
|
|
|
sharding_spec_checklist = []
|
|
|
|
|
for strategy in input_node.strategies_vector:
|
|
|
|
|
# It looks a little bit confusing, the input of the processing node
|
|
|
|
|
# is the output of the input_node.
|
|
|
|
|
input_sharding_spec = strategy.output_sharding_spec
|
|
|
|
|
assert isinstance(input_sharding_spec,
|
|
|
|
|
ShardingSpec), f'The input node should NOT be a tuple of tensor.'
|
|
|
|
|
if input_sharding_spec in sharding_spec_checklist:
|
|
|
|
|
continue
|
|
|
|
|
sharding_spec_checklist.append(input_sharding_spec)
|
|
|
|
|
dim_partition_dict = deepcopy(input_sharding_spec.dim_partition_dict)
|
2022-09-13 07:43:22 +00:00
|
|
|
output_sharding_spec = generate_sharding_spec(node, self.device_mesh, dim_partition_dict)
|
2022-08-30 08:32:09 +00:00
|
|
|
name = f'{input_sharding_spec.sharding_sequence} -> {output_sharding_spec.sharding_sequence}'
|
|
|
|
|
# TODO: use meta_info_prop to profile memory cost and compute cost
|
|
|
|
|
compute_cost = node._meta_data.numel()
|
|
|
|
|
memory_cost = 0
|
|
|
|
|
|
2022-09-13 10:30:18 +00:00
|
|
|
resharding_costs = generate_resharding_costs(strategies_vector.predecessor_nodes,
|
|
|
|
|
[input_sharding_spec])
|
2022-08-30 08:32:09 +00:00
|
|
|
|
|
|
|
|
# to prevent the resharding happening, set their resharding cost to inf.
|
|
|
|
|
resharding_costs[input_node] = [
|
|
|
|
|
0 if cost == 0 else math.inf for cost in resharding_costs[input_node]
|
|
|
|
|
]
|
|
|
|
|
sharding_strategy = ShardingStrategy(name,
|
|
|
|
|
output_sharding_spec,
|
|
|
|
|
compute_cost=compute_cost,
|
|
|
|
|
memory_cost=memory_cost,
|
|
|
|
|
resharding_costs=resharding_costs,
|
|
|
|
|
input_shardings=[input_sharding_spec])
|
|
|
|
|
strategies_vector.append(sharding_strategy)
|
|
|
|
|
|
|
|
|
|
# torch.var_mean
|
|
|
|
|
elif target == torch.var_mean:
|
|
|
|
|
dim = node.kwargs['dim']
|
|
|
|
|
input_tensor_node = strategies_vector.predecessor_nodes[0]
|
|
|
|
|
for strategy in input_tensor_node.strategies_vector:
|
|
|
|
|
input_sharding_spec = strategy.output_sharding_spec
|
|
|
|
|
assert isinstance(input_sharding_spec,
|
|
|
|
|
ShardingSpec), f'The input node should NOT be a tuple of tensor.'
|
|
|
|
|
entire_shape_input = input_sharding_spec.entire_shape
|
|
|
|
|
dim_partition_dict_input = input_sharding_spec.dim_partition_dict
|
|
|
|
|
name = f'{new_input_sharding_spec.sharding_sequence} -> ({output_sharding_spec.sharding_sequence}, {output_sharding_spec.sharding_sequence})'
|
|
|
|
|
if dim in dim_partition_dict_input:
|
|
|
|
|
# We need to make the action dimension in replicate status
|
|
|
|
|
dim_partition_dict_for_input = deepcopy(dim_partition_dict_input)
|
|
|
|
|
dim_partition_dict_for_input.pop(dim)
|
|
|
|
|
new_input_sharding_spec = ShardingSpec(self.device_mesh,
|
|
|
|
|
entire_shape_input,
|
|
|
|
|
dim_partition_dict=dim_partition_dict_for_input)
|
|
|
|
|
entire_shape_output = deepcopy(entire_shape_input)
|
|
|
|
|
entire_shape_output.pop(dim)
|
|
|
|
|
dim_partition_dict_for_output = deepcopy(dim_partition_dict_for_input)
|
|
|
|
|
output_sharding_spec = ShardingSpec(self.device_mesh,
|
|
|
|
|
entire_shape_output,
|
|
|
|
|
dim_partition_dict=dim_partition_dict_for_input)
|
|
|
|
|
# TODO: use meta_info_prop to profile origin memory cost and compute cost, then divide them depending on sharding spec.
|
|
|
|
|
compute_cost = 0
|
|
|
|
|
memory_cost = 0
|
2022-09-13 10:30:18 +00:00
|
|
|
resharding_costs = generate_resharding_costs(strategies_vector.predecessor_nodes,
|
|
|
|
|
[new_input_sharding_spec])
|
2022-08-30 08:32:09 +00:00
|
|
|
sharding_strategy = ShardingStrategy(name, (output_sharding_spec, output_sharding_spec),
|
|
|
|
|
compute_cost=compute_cost,
|
|
|
|
|
memory_cost=memory_cost,
|
|
|
|
|
resharding_costs=resharding_costs,
|
|
|
|
|
input_shardings=[new_input_sharding_spec])
|
|
|
|
|
|
|
|
|
|
else:
|
|
|
|
|
entire_shape_output = deepcopy(entire_shape_input)
|
|
|
|
|
entire_shape_output.pop(dim)
|
|
|
|
|
dim_partition_dict_for_output = deepcopy(dim_partition_dict_input)
|
|
|
|
|
output_sharding_spec = ShardingSpec(self.device_mesh,
|
|
|
|
|
entire_shape_output,
|
|
|
|
|
dim_partion_dict=dim_partition_dict_input)
|
|
|
|
|
# TODO: use meta_info_prop to profile origin memory cost and compute cost, then divide them depending on sharding spec.
|
|
|
|
|
compute_cost = 0
|
|
|
|
|
memory_cost = 0
|
2022-09-13 10:30:18 +00:00
|
|
|
resharding_costs = generate_resharding_costs(strategies_vector.predecessor_nodes,
|
|
|
|
|
[input_sharding_spec])
|
2022-08-30 08:32:09 +00:00
|
|
|
sharding_strategy = ShardingStrategy(name, (output_sharding_spec, output_sharding_spec),
|
|
|
|
|
compute_cost=compute_cost,
|
|
|
|
|
memory_cost=memory_cost,
|
|
|
|
|
resharding_costs=resharding_costs,
|
|
|
|
|
input_shardings=[input_sharding_spec])
|
|
|
|
|
|
|
|
|
|
strategies_vector.append(sharding_strategy)
|
|
|
|
|
|
|
|
|
|
# operator.getitem
|
|
|
|
|
elif target == operator.getitem:
|
|
|
|
|
index = node.args[1]
|
|
|
|
|
input_tensor_node = strategies_vector.predecessor_nodes[0]
|
|
|
|
|
for strategy in input_tensor_node.strategies_vector:
|
|
|
|
|
input_sharding_spec = input_tensor_node.output_sharding_spec[index]
|
|
|
|
|
assert isinstance(input_sharding_spec, ShardingSpec), f'This assertion is used to debug.'
|
|
|
|
|
dim_partition_dict_for_output = deepcopy(input_sharding_spec.dim_partition_dict)
|
|
|
|
|
entire_shape_output = deepcopy(input_sharding_spec.entire_shape)
|
|
|
|
|
output_sharding_spec = ShardingSpec(self.device_mesh,
|
|
|
|
|
entire_shape_output,
|
|
|
|
|
dim_partition_dict=dim_partition_dict_for_output)
|
|
|
|
|
# TODO: use meta_info_prop to profile origin memory cost and compute cost, then divide them depending on sharding spec.
|
|
|
|
|
compute_cost = 0
|
|
|
|
|
memory_cost = 0
|
2022-09-13 10:30:18 +00:00
|
|
|
resharding_costs = generate_resharding_costs(strategies_vector.predecessor_nodes,
|
|
|
|
|
[input_sharding_spec])
|
2022-08-30 08:32:09 +00:00
|
|
|
# to prevent the resharding happening, set their resharding cost to inf.
|
|
|
|
|
resharding_costs[input_tensor_node] = [
|
|
|
|
|
cost if cost == 0 else math.inf for cost in resharding_costs[input_tensor_node]
|
|
|
|
|
]
|
|
|
|
|
sharding_strategy = ShardingStrategy(name,
|
|
|
|
|
output_sharding_spec,
|
|
|
|
|
compute_cost=compute_cost,
|
|
|
|
|
memory_cost=memory_cost,
|
|
|
|
|
resharding_costs=resharding_costs,
|
|
|
|
|
input_shardings=[input_tensor_node.output_sharding_spec])
|
|
|
|
|
strategies_vector.append(sharding_strategy)
|
|
|
|
|
|
|
|
|
|
# other function
|
|
|
|
|
else:
|
|
|
|
|
raise RuntimeError(f'{target} function is NOT supported now.')
|
|
|
|
|
|
|
|
|
|
# output node
|
|
|
|
|
if node.op == 'output':
|
2022-09-13 06:47:09 +00:00
|
|
|
if self.solver_options.fast:
|
2022-08-30 08:32:09 +00:00
|
|
|
# create sharding strategy for output
|
|
|
|
|
name = 'Replica Output'
|
|
|
|
|
input_nodes = strategies_vector.predecessor_nodes
|
|
|
|
|
input_sharding_specs = []
|
|
|
|
|
for input_node in input_nodes:
|
|
|
|
|
dim_partition_dict_for_input = {}
|
|
|
|
|
entire_shape = input_node._meta_data.shape
|
|
|
|
|
sharding_spec = ShardingSpec(self.device_mesh,
|
|
|
|
|
entire_shape,
|
|
|
|
|
dim_partition_dict=dim_partition_dict_for_input)
|
|
|
|
|
input_sharding_specs.append(sharding_spec)
|
|
|
|
|
|
|
|
|
|
dim_partition_dict = {}
|
|
|
|
|
output_sharding_spec = input_sharding_specs
|
|
|
|
|
# TODO: use meta_info_prop to profile memory cost
|
|
|
|
|
memory_cost = 0
|
2022-09-13 10:30:18 +00:00
|
|
|
resharding_costs = generate_resharding_costs(strategies_vector.predecessor_nodes,
|
|
|
|
|
input_sharding_specs)
|
2022-09-13 04:07:09 +00:00
|
|
|
|
|
|
|
|
# clear the resharding cost for the output node
|
|
|
|
|
# TODO: we may remove this in final version
|
|
|
|
|
if True:
|
|
|
|
|
for prev_node, resharding_cost_list in resharding_costs.items():
|
|
|
|
|
resharding_costs[prev_node] = [0] * len(resharding_cost_list)
|
|
|
|
|
|
2022-08-30 08:32:09 +00:00
|
|
|
sharding_strategy_attribute = ShardingStrategy(name,
|
|
|
|
|
output_sharding_spec,
|
|
|
|
|
memory_cost=memory_cost,
|
|
|
|
|
resharding_costs=resharding_costs)
|
|
|
|
|
strategies_vector.append(sharding_strategy_attribute)
|
|
|
|
|
|
|
|
|
|
self.remove_duplicated_strategy(strategies_vector)
|
|
|
|
|
setattr(node, 'strategies_vector', strategies_vector)
|
|
|
|
|
self.leaf_strategies.append(strategies_vector)
|
|
|
|
|
self.strategy_map[node] = strategies_vector
|
