ddp.py

import os

import torch
import torch.distributed as dist
import torch.multiprocessing as mp
import torch.nn as nn
import torch.optim as optim
from torch.nn.parallel import DistributedDataParallel as DDP


def setup(rank, world_size):
    os.environ['MASTER_ADDR'] = 'localhost'
    os.environ['MASTER_PORT'] = '12355'

    # initialize the process group
    dist.init_process_group("gloo", rank=rank, world_size=world_size)


def cleanup():
    dist.destroy_process_group()


class ToyModel(nn.Module):
    def __init__(self):
        super(ToyModel, self).__init__()
        self.net1 = nn.Linear(10000, 100)
        self.relu = nn.ReLU()
        self.net2 = nn.Linear(100, 5)

    def forward(self, x):
        return self.net2(self.relu(self.net1(x)))


def demo_basic(rank, world_size):
    print(f"Running basic DDP example on rank {rank}.")
    setup(rank, world_size)

    # create model and move it to GPU with id rank
    model = ToyModel().to(rank)
    ddp_model = DDP(model, device_ids=[rank])

    loss_fn = nn.MSELoss()
    optimizer = optim.SGD(ddp_model.parameters(), lr=0.001)

    for epoch in range(1000):
        # print(f"epoch {epoch}")
        optimizer.zero_grad()
        outputs = ddp_model(torch.randn(20, 10000))
        labels = torch.randn(20, 5).to(rank)
        loss_fn(outputs, labels).backward()
        optimizer.step()

    cleanup()


def run_demo(demo_fn, world_size):
    mp.spawn(demo_fn,
             args=(world_size,),
             nprocs=world_size,
             join=True)


if __name__ == "__main__":
    size = 2
    run_demo(demo_basic, size)