[shardformer] integrated linear 1D with dtensor (#3996)

* [shardformer] integrated linear 1D with dtensor

* polish code

colossalai/nn/layer/base_layer.py

@@ -10,6 +10,7 @@ from colossalai.core import global_context as gpc
 
 
 class ParallelLayer(nn.Module):
+
     global_state_dict: bool = True
 
     def __init__(self):

colossalai/shardformer/layer/_operation.py

@@ -54,10 +54,10 @@ class LinearWithAsyncCommunication(torch.autograd.Function):
     """
 
     @staticmethod
-    def forward(ctx, input_, weight, bias, parallel_mode, async_grad_allreduce):
+    def forward(ctx, input_, weight, bias, process_group, async_grad_allreduce):
         ctx.save_for_backward(input_, weight)
         ctx.use_bias = bias is not None
-        ctx.parallel_mode = parallel_mode
+        ctx.process_group = process_group
         ctx.async_grad_allreduce = async_grad_allreduce
 
         output = torch.matmul(input_, weight.t())
@@ -74,12 +74,13 @@ class LinearWithAsyncCommunication(torch.autograd.Function):
         grad_input = grad_output.matmul(weight)
         grad_output = grad_output.contiguous()
         # Convert the tensor shapes to 2D for execution compatibility
-        grad_output = grad_output.view(grad_output.shape[0] * grad_output.shape[1], grad_output.shape[2])
-        total_input = total_input.view(total_input.shape[0] * total_input.shape[1], total_input.shape[2])
+        if len(grad_output.shape) > 2:
+            grad_output = grad_output.view(-1, grad_output.shape[-1])
+            total_input = total_input.view(-1, total_input.shape[-1])
 
         if ctx.async_grad_allreduce:
             # Asynchronous all-reduce
-            handle = dist.all_reduce(grad_input, group=gpc.get_group(ctx.parallel_mode), async_op=True)
+            handle = dist.all_reduce(grad_input, group=ctx.process_group, async_op=True)
             # Delay the start of weight gradient computation shortly (3us) to have
             # all-reduce scheduled first and have GPU resources allocated
             _ = torch.empty(1, device=grad_output.device) + 1
@@ -93,5 +94,123 @@ class LinearWithAsyncCommunication(torch.autograd.Function):
         return grad_input, grad_weight, grad_bias, None, None, None
 
 
-def linear_with_async_comm(input_, weight, bias, parallel_mode, async_grad_allreduce):
-    return LinearWithAsyncCommunication.apply(input_, weight, bias, parallel_mode, async_grad_allreduce)
+class _SplitForwardGatherBackward(torch.autograd.Function):
+    """
+    Split the input and keep only the corresponding chuck to the rank.
+
+    Args:
+        input_ (`torch.Tensor`): input matrix.
+        dim (int): the dimension to perform split and gather
+        process_group (`torch.distributed.ProcessGroup`): the process group used for collective communication
+
+    """
+
+    @staticmethod
+    def forward(ctx, input_, dim, process_group):
+        ctx.process_group = process_group
+        ctx.dim = dim
+        return _split(input_, dim, process_group)
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        return _gather(grad_output, ctx.dim, ctx.process_group), None, None
+
+
+class _ReduceInput(torch.autograd.Function):
+    """
+    All-reduce the input from the model parallel region.
+
+    Args:
+        input_: input matrix.
+        parallel_mode: parallel mode.
+    """
+
+    @staticmethod
+    def forward(ctx, input_, process_group):
+        return _reduce(input_, process_group)
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        return grad_output, None
+
+
+def _reduce(input_, process_group):
+    # skip if only one rank involved
+    if dist.get_world_size(process_group) == 1:
+        return input_
+    else:
+        dist.all_reduce(input_, group=process_group)
+        return input_
+
+
+def _split(input_, dim=-1, process_group=None):
+    # skip if only one rank involved
+    world_size = dist.get_world_size(process_group)
+    if world_size == 1:
+        return input_
+
+    # Split along last dimension.
+    dim_size = input_.size(dim)
+    assert dim_size % world_size == 0, \
+        f'The dimension to split ({dim_size}) is not a multiple of world size ({world_size}), ' \
+        f'cannot split tensor evenly'
+
+    tensor_list = torch.split(input_, dim_size // world_size, dim=dim)
+    rank = dist.get_rank(process_group)
+    output = tensor_list[rank].contiguous()
+
+    return output
+
+
+def _gather(input_, dim=-1, process_group=None):
+    # skip if only one rank involved
+    world_size = dist.get_world_size(process_group)
+    if world_size == 1:
+        return input_
+
+    # all gather
+    rank = dist.get_rank(process_group)
+    tensor_list = [torch.empty_like(input_) for _ in range(world_size)]
+    tensor_list[rank] = input_
+    torch.distributed.all_gather(tensor_list, input_, group=process_group)
+
+    # concat
+    output = torch.cat(tensor_list, dim=dim).contiguous()
+
+    return output
+
+
+class _GatherForwardSplitBackward(torch.autograd.Function):
+    """Gather the input from model parallel region and concatenate.
+
+    Args:
+        input_: input matrix.
+        parallel_mode: parallel mode.
+        dim: dimension
+    """
+
+    @staticmethod
+    def forward(ctx, input_, dim, process_group):
+        ctx.process_group = process_group
+        ctx.dim = dim
+        return _gather(input_, dim, process_group)
+
+    @staticmethod
+    def backward(ctx, grad_output):
+        return _split(grad_output, ctx.dim, ctx.process_group), None, None
+
+
+def linear_with_async_comm(input_, weight, bias, process_group, async_grad_allreduce):
+    return LinearWithAsyncCommunication.apply(input_, weight, bias, process_group, async_grad_allreduce)
+
+
+def gather_forward_split_backward(input_, dim, process_group):
+    return _GatherForwardSplitBackward.apply(input_, dim, process_group)
+
+
+def split_forward_gather_backward(input_, dim, process_group):
+    return _SplitForwardGatherBackward.apply(input_, dim, process_group)
+
+
+def reduce_input(input_, process_group):
+    return _ReduceInput.apply(input_, process_group)

colossalai/shardformer/layer/dropout.py

-import os
-from contextlib import contextmanager
-
 import torch
+import torch.distributed as dist
 import torch.nn as nn
 
-
-class SeedManager:
-    """
-    This class is a random state manager to change random state for different random seed.
-
-    """
-
-    def __init__(self):
-        original_state = torch.cuda.get_rng_state()
-        # TODO: unify this seed manager with the colossalai.context.random
-        seed = os.getpid()
-        torch.cuda.manual_seed(int(seed))
-        self.dropout_state = torch.cuda.get_rng_state()
-        torch.cuda.set_rng_state(original_state)
-
-    def set_mode(self, rng_state):
-        torch.cuda.set_rng_state(rng_state)
-
-    def get_current_mode(self):
-        current_state = torch.cuda.get_rng_state()
-        return current_state
-
-    @contextmanager
-    def dropout_mode(self):
-        """
-        This is a context manager to change the dropout state and recover the original state.
-
-        Usage:
-        ::
-            >>> with _seed_manager.dropout_mode():
-            >>>     input = super().forward(input)
-        """
-        try:
-            current_mode = self.get_current_mode()
-            yield self.set_mode(self.dropout_state)
-        finally:
-            self.dropout_state = self.get_current_mode()
-            self.set_mode(current_mode)
-
-
-_seed_manager = SeedManager()
+from .utils import create_randomizer_with_offset
 
 
 class Dropout1D(nn.Dropout):
 
-    def __init__(self, p=0.5, inplace=False):
+    def __init__(self, p=0.5, inplace=False, process_group=None):
         super().__init__(p, inplace)
 
+        # offset the seed with randomizer index and rank
+        seed = torch.random.initial_seed()
+        self.randomizer = create_randomizer_with_offset(seed, process_group=process_group)
+
     def forward(self, input):
-        with _seed_manager.dropout_mode():
+        with self.randomizer.fork_rng():
             input = super().forward(input)
         return input

colossalai/shardformer/layer/utils.py

+from contextlib import contextmanager
+
+import torch
+import torch.distributed as dist
+from torch.distributed import ProcessGroup
+
+
+class Randomizer:
+    """
+    Randomizer enables the program to be executed under a different seed within the context.
+
+    Example:
+
+    ```python
+    randomizer = Randomizer(seed=1024)
+
+    with randomizer.fork():
+        # do something here with seed 1024
+        do_something()
+    ```
+
+    Args:
+        seed (int): The random seed to set.
+        enable_cpu (bool): fork the CPU RNG state as well.
+        with_index (bool): whether to use the index of the randomizer.
+    """
+
+    _INDEX = 0
+
+    def __init__(self, seed: int):
+        # TODO: remove colossalai.context.random
+
+        self.seed = seed
+
+        # Handle CUDA rng state
+        # 1. get the current rng state
+        # 2. set the seed and store the rng state
+        # 3. recover the original rng state
+        cuda_original_rng_state = torch.cuda.get_rng_state()
+        torch.cuda.manual_seed(seed)
+        self.cuda_rng_state = torch.cuda.get_rng_state()
+        torch.cuda.set_rng_state(cuda_original_rng_state)
+
+        # to the same for cpu rng state
+        cpu_original_rng_state = torch.get_rng_state()
+        torch.manual_seed(seed)
+        self.cpu_rng_state = torch.get_rng_state()
+        torch.set_rng_state(cpu_original_rng_state)
+
+    def _set_cuda_rng_state(self, rng_state):
+        torch.cuda.set_rng_state(rng_state)
+
+    def _get_cuda_rng_state(self):
+        current_state = torch.cuda.get_rng_state()
+        return current_state
+
+    def _set_cpu_rng_state(self, rng_state):
+        torch.set_rng_state(rng_state)
+
+    def _get_cpu_rng_state(self):
+        current_state = torch.get_rng_state()
+        return current_state
+
+    @contextmanager
+    def fork_rng(self, enable_cpu: bool = False):
+        """
+        This is a context manager to change the dropout state and recover the original state.
+
+        Usage:
+        ::
+            >>> with _seed_manager.dropout_mode():
+            >>>     input = super().forward(input)
+        """
+        try:
+            current_cuda_rng_state = self._get_cuda_rng_state()
+            self._set_cuda_rng_state(self.cuda_rng_state)
+
+            if enable_cpu:
+                current_cpu_rng_state = self._get_cpu_rng_state()
+                self._set_cpu_rng_state(self.cpu_rng_state)
+            yield
+        finally:
+            self.cuda_rng_state = self._get_cuda_rng_state()
+            self._set_cuda_rng_state(current_cuda_rng_state)
+
+            if enable_cpu:
+                self.cpu_rng_state = self._get_cpu_rng_state()
+                self._set_cpu_rng_state(current_cpu_rng_state)
+
+    @staticmethod
+    def index():
+        """
+        Return the index of the randomizer. The index is useful when the user wants
+        to introduce some randomness in the program.
+
+        Note:
+        The index will increment by one each time this method is called.
+
+        Example:
+
+        ```python
+        # assume we need a randomizer to init the weight of different layers
+        # we can use the index of the randomizer to do so that
+        # each layer has its own randomizer with a different seed
+        base_seed = torch.random.initial_seed()
+        seed = base_seed + Randomizer.index()
+        randomizer = Randomizer(seed)
+
+        with randomizer.fork():
+            init_weights()
+        ```
+
+        """
+        idx = Randomizer._INDEX
+        Randomizer._INDEX += 1
+        return idx
+
+
+def create_randomizer_with_offset(seed: int, process_group: ProcessGroup = None):
+    """
+    Create a randomizer with an offset. The offset is equal to the rank of the process and the index of the randomizer.
+
+    Args:
+        seed (int): The base random seed to set.
+        enable_cpu (bool): fork the CPU RNG state as well.
+        process_group (ProcessGroup): the process group to get the rank from.
+
+    Returns:
+        Randomizer: the randomizer with offset.
+    """
+    offset = Randomizer.index()
+
+    if dist.is_initialized():
+        rank = dist.get_rank(process_group)
+        offset += rank
+
+    seed += offset
+    return Randomizer(seed=seed)

colossalai/tensor/d_tensor/api.py

+from typing import Union
+
+import torch
+import torch.distributed as dist
+from torch.distributed import ProcessGroup
+
+from colossalai.device.device_mesh import DeviceMesh
+
+from .d_tensor import DTensor
+from .sharding_spec import ShardingSpec
+
+
+def shard_rowwise(tensor: torch.Tensor, group_or_device_mesh: Union[ProcessGroup, DeviceMesh] = None) -> DTensor:
+    """
+    Shard the first dim of the given tensor
+    """
+    # if the group_or_device_mesh is None, we shard the tensor with respect to the global process group
+    if group_or_device_mesh is None:
+        group_or_device_mesh = dist.GroupMember.WORLD
+
+    if isinstance(group_or_device_mesh, ProcessGroup):
+        device_mesh = DeviceMesh.from_process_group(group_or_device_mesh)
+    else:
+        assert len(group_or_device_mesh.shape) == 1, 'Only 1D DeviceMesh is accepted for row-wise sharding.'
+        device_mesh = group_or_device_mesh
+    sharding_spec = ShardingSpec(dim_size=tensor.dim(), dim_partition_dict={0: [0]})
+    return DTensor(tensor, device_mesh, sharding_spec)
+
+
+def shard_colwise(tensor: torch.Tensor, group_or_device_mesh: Union[ProcessGroup, DeviceMesh] = None) -> DTensor:
+    """
+    Shard the first dim of the given tensor
+    """
+    # if the group_or_device_mesh is None, we shard the tensor with respect to the global process group
+    if group_or_device_mesh is None:
+        group_or_device_mesh = dist.GroupMember.WORLD
+
+    if isinstance(group_or_device_mesh, ProcessGroup):
+        device_mesh = DeviceMesh.from_process_group(group_or_device_mesh)
+    else:
+        assert len(group_or_device_mesh.shape) == 1, 'Only 1D DeviceMesh is accepted for row-wise sharding.'
+        device_mesh = group_or_device_mesh
+    sharding_spec = ShardingSpec(dim_size=tensor.dim(), dim_partition_dict={-1: [0]})
+    return DTensor(tensor, device_mesh, sharding_spec)

colossalai/tensor/d_tensor/layout.py

@@ -34,7 +34,7 @@ class Layout:
     def get_sharded_shape_per_device(self):
         sharded_shape = list(self.entire_shape)
         for dim, shard_list in self.sharding_spec.dim_partition_dict.items():
-            mesh_list = [self.device_mesh.mesh_shape[mesh_dim] for mesh_dim in shard_list]
+            mesh_list = [self.device_mesh.shape[mesh_dim] for mesh_dim in shard_list]
             shard_partitions = reduce(operator.mul, mesh_list, 1)
             assert sharded_shape[
                 dim] % shard_partitions == 0, f'Cannot shard dimension {dim} into {shard_partitions} partitions.'
@@ -45,14 +45,15 @@ class Layout:
         sharding_spec = self.sharding_spec
 
         # make sure all axes in logical device mesh only be used once
-        dim_check_list = list(range(self.device_mesh.logical_mesh_id.dim()))
-        for dim, shard_list in sharding_spec.dim_partition_dict.items():
-            for element in shard_list:
-                if element in dim_check_list:
-                    dim_check_list.remove(element)
-                else:
-                    raise DuplicatedShardingDimensionError(
-                        f"find an invalid sharding axis {element} in dim_partition_dict in tensor dimension {dim}.")
+        if self.device_mesh.logical_mesh_id is not None:
+            dim_check_list = list(range(self.device_mesh.logical_mesh_id.dim()))
+            for dim, shard_list in sharding_spec.dim_partition_dict.items():
+                for element in shard_list:
+                    if element in dim_check_list:
+                        dim_check_list.remove(element)
+                    else:
+                        raise DuplicatedShardingDimensionError(
+                            f"find an invalid sharding axis {element} in dim_partition_dict in tensor dimension {dim}.")
 
         # make sure that the sharding for a dimension is divisible by the number of devices
         for dim, shard_list in sharding_spec.dim_partition_dict.items():
@@ -60,7 +61,7 @@ class Layout:
             num_devices = 1
 
             for element in shard_list:
-                num_devices *= self.device_mesh.mesh_shape[element]
+                num_devices *= self.device_mesh.shape[element]
 
             if tensor_dim_size % num_devices != 0:
                 raise ShardingNotDivisibleError(

colossalai/tensor/d_tensor/layout_converter.py

@@ -304,7 +304,7 @@ class LayoutConverter(metaclass=SingletonMeta):
         process_groups_dict = source_layout.device_mesh.process_groups_dict
 
         # legal sharding dims means the mesh_id is still available to use.
-        legal_sharding_dims = [i for i in range(len(source_layout.device_mesh.mesh_shape))]
+        legal_sharding_dims = [i for i in range(len(source_layout.device_mesh.shape))]
         for dim, shard_list in source_spec.dim_partition_dict.items():
             for element in shard_list:
                 legal_sharding_dims.remove(element)

tests/test_shardformer/test_layer/test_linear_1d.py

+import torch
+import torch.distributed as dist
+import torch.nn as nn
+from torch.testing import assert_close
+
+import colossalai
+from colossalai.shardformer.layer.layers import Linear1D_Col, Linear1D_Row
+from colossalai.testing import parameterize, rerun_if_address_is_in_use, spawn
+
+
+def check_linear_1d_col():
+    linear = nn.Linear(32, 128).cuda()
+    linear_col = Linear1D_Col.from_native_module(linear, process_group=None, gather_output=True)
+
+    assert linear_col.weight.shape == torch.Size([64, 32])
+    assert linear_col.bias.shape == torch.Size([64])
+
+    # check computation correctness
+    x = torch.rand(4, 32).cuda()
+    out = linear(x)
+    gather_out = linear_col(x)
+    assert_close(out, gather_out)
+
+    # check backward correctness
+    out.sum().backward()
+    gather_out.sum().backward()
+
+    rank = dist.get_rank()
+    target_grad = torch.chunk(linear.weight.grad, 2, dim=0)[rank]
+    assert_close(target_grad, linear_col.weight.grad)
+
+
+def check_linear_1d_row():
+    linear = nn.Linear(32, 128).cuda()
+    linear_row = Linear1D_Row.from_native_module(linear, process_group=None, parallel_input=False)
+
+    assert linear_row.weight.shape == torch.Size([128, 16])
+    assert linear_row.bias.shape == torch.Size([128])
+
+    # check computation correctness
+    x = torch.rand(4, 32).cuda()
+    out = linear(x)
+    gather_out = linear_row(x)
+    assert_close(out, gather_out)
+
+    # check backward correctness
+    out.sum().backward()
+    gather_out.sum().backward()
+
+    rank = dist.get_rank()
+    target_grad = torch.chunk(linear.weight.grad, 2, dim=1)[rank]
+    assert_close(target_grad, linear_row.weight.grad)
+
+
+def run_dist(rank, world_size, port):
+    colossalai.launch(config={}, rank=rank, world_size=world_size, host='localhost', port=port, backend='nccl')
+    check_linear_1d_col()
+    check_linear_1d_row()
+
+
+@rerun_if_address_is_in_use()
+def test_linear():
+    spawn(run_dist, nprocs=2)
+
+
+if __name__ == '__main__':
+    test_linear()