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/tensor/d_tensor
Edenzzzz 43995ee436
[Feature] Distributed optimizers: Lamb, Galore, CAME and Adafactor (#5694)
6 months ago
..
README.md [misc] refactor launch API and tensor constructor (#5666) 7 months ago
__init__.py [Feature] Distributed optimizers: Lamb, Galore, CAME and Adafactor (#5694) 6 months ago
api.py [Feature] Distributed optimizers: Lamb, Galore, CAME and Adafactor (#5694) 6 months ago
comm_spec.py [hotfix] fixed memory usage of shardformer module replacement (#5122) 1 year ago
layout.py [misc] update pre-commit and run all files (#4752) 1 year ago
layout_converter.py [shardformer] refactor embedding resize (#5603) 7 months ago
misc.py
sharding_spec.py [Feature] Distributed optimizers: Lamb, Galore, CAME and Adafactor (#5694) 6 months ago
utils.py [misc] update pre-commit and run all files (#4752) 1 year ago

README.md

🔢 Distributed Tensor

📚 Table of Contents

🔗 Introduction

Distributed tensor is a type of tensor that is distributed across multiple devices. It is a wrapper of PyTorch tensor, and it is used to support distributed training. It can represent the device topology and tensor placement over the devices in the topology. It also provides a set of APIs to manipulate the distributed tensor.

📝 Design

Our implementation is inspired by the work Alpa, which unifies data parallelism and tensor parallelism as intra-op parallelism. It uses notations S to represent the sharded dimension and R to represent the replicated dimension. For example, given a 2D matrix, [S, R] represents the tensor is sharded over the first dimension.

Each sharded dimension will have a subscript to represent its placement over the devices. Assuming we have 4 GPUs and the GPUs are arranged in a 2 x 2 manner. Let's say we have a 2D matrix like below:

    [1,  2,  3,  4 ]
A = [4,  5,  6,  7 ]
    [8,  9,  10, 11]
    [12, 13, 14, 15]

[S0, R] would mean that the first dimension is sharded over the rows in the device topology.

| --------------------—————————————————————-|
|                     |                     |
|  [1,  2,  3,  4 ]   |  [1,  2,  3,  4 ]   |
|  [4,  5,  6,  7 ]   |  [4,  5,  6,  7 ]   |
|                     |                     |
| --------------------——————————————————-----
|                     |                     |
|  [8,  9,  10, 11]   |  [8,  9,  10, 11]   |
|  [12, 13, 14, 15]   |  [12, 13, 14, 15]   |
|                     |                     |
| --------------------——————————————————-----

[S01, R] would mean that the first dimension is sharded over both the row and column in the device topology.

| --------------------—————————————————————-|
|                     |                     |
|  [1,  2,  3,  4 ]   |  [4,  5,  6,  7 ]   |
|                     |                     |
| --------------------——————————————————-----
|                     |                     |
|  [8,  9,  10, 11]   |  [12, 13, 14, 15]   |
|                     |                     |
| --------------------——————————————————-----

🔨 Usage

A sample API usage is given below.

import torch

import colossalai
from colossalai.device.device_mesh import DeviceMesh
from colossalai.tensor.d_tensor import DTensor, ShardingSpec

colossalai.launch_from_torch()

# define your device mesh
# assume you have 4 GPUs
physical_mesh_id = torch.arange(0, 4)
mesh_shape = (2, 2)
device_mesh = DeviceMesh(physical_mesh_id, mesh_shape, init_process_group=True)

# define a tensor
a = torch.rand(16, 32).cuda()

# create sharding spec for the tensor
# assume the sharding spec is [S0, R]
dim_partition_dict = {0: [0]}
sharding_spec = ShardingSpec(a.dim(), dim_partition_dict)

# create a distributed tensor
d_tensor = DTensor(a, device_mesh, sharding_spec)
print(d_tensor)

global_tensor = d_tensor.to_global()
print(global_tensor)

🎈 Progress Log

  • Support layout conversion
  • Support sharding on 2D device mesh
  • Support sharding on 3D device mesh
  • Support sharding 4D device mesh
  • Support sharding info saving and offline tensor merge (we can save tensor as dtensor and gather the tensors back to the global tensor based on the sharding info in a single process in CPU, useful for distributed training checkpoint load and save.)