DistributedDataParallel in PyTorch - Cheat Sheet & Quick Revision

Recall & Review

beginner

What is DistributedDataParallel in PyTorch?

DistributedDataParallel (DDP) is a PyTorch module that helps train models on multiple GPUs or machines by splitting data and synchronizing gradients efficiently.

intermediate

Why use DistributedDataParallel instead of DataParallel?

DistributedDataParallel is faster and more scalable because it uses multiple processes and synchronizes gradients in parallel, while DataParallel uses a single process and can be slower.

intermediate

How does DistributedDataParallel synchronize model updates?

DDP synchronizes gradients by averaging them across all processes after each backward pass, ensuring all model copies stay consistent during training.

beginner

What is the role of torch.distributed.init_process_group() in DDP?

It initializes the communication between processes, setting up the environment so that DistributedDataParallel can coordinate training across GPUs or machines.

beginner

What must you do with your dataset when using DistributedDataParallel?

You should use a DistributedSampler to split the dataset so each process gets a unique subset, avoiding overlap and ensuring efficient training.

What does DistributedDataParallel primarily help with?

A Visualizing model predictions

B Reducing model size

C Training models on multiple GPUs or machines efficiently

D Improving data loading speed

DistributedDataParallel is designed to train models efficiently across multiple GPUs or machines by distributing data and synchronizing gradients.

Which function initializes communication between processes in DDP?

A torch.distributed.init_process_group()

B torch.nn.Module()

C torch.optim.SGD()

D torch.utils.data.DataLoader()

torch.distributed.init_process_group() sets up the communication needed for DistributedDataParallel to work.

What is the purpose of DistributedSampler in DDP?

A To shuffle data randomly

B To split dataset uniquely across processes

C To increase batch size

D To reduce model parameters

DistributedSampler ensures each process gets a unique subset of data, avoiding overlap during distributed training.

How does DDP synchronize model updates?

A By copying weights from one GPU to another

B By saving checkpoints

C By sending data to CPU

D By averaging gradients across all processes after backward pass

DDP averages gradients from all processes to keep model copies consistent.

Compared to DataParallel, DistributedDataParallel is:

A Faster and more scalable

B Only for CPU training

C Slower and less scalable

D Used for model evaluation only

DistributedDataParallel uses multiple processes and parallel communication, making it faster and more scalable than DataParallel.

Explain how DistributedDataParallel works to train a model on multiple GPUs.

Think about how data and model updates are shared across GPUs.

Describe the steps needed to prepare your PyTorch training code to use DistributedDataParallel.

Focus on setup and data handling for distributed training.

Practice

1. What is the main purpose of using DistributedDataParallel in PyTorch?

easy

A. To automatically tune hyperparameters during training

B. To reduce the size of the model for deployment

C. To convert a model to run only on CPU

D. To speed up training by using multiple GPUs or machines simultaneously

Solution

1. Step 1: Understand the role of DistributedDataParallel

   DistributedDataParallel (DDP) is designed to run training across multiple GPUs or machines to speed up the process.

2. Step 2: Compare options with DDP's purpose

   Options B, C, and D do not relate to parallel training or speed-up. Only To speed up training by using multiple GPUs or machines simultaneously correctly describes DDP's main purpose.

3. Final Answer:

   To speed up training by using multiple GPUs or machines simultaneously -> Option D

4. Quick Check:

   DDP speeds training = A [OK]

Hint: DDP means training faster with many GPUs [OK]

Common Mistakes:

• Thinking DDP reduces model size
• Confusing DDP with CPU-only training
• Assuming DDP tunes hyperparameters automatically

2. Which of the following is the correct way to wrap a PyTorch model with DistributedDataParallel assuming model is your neural network and device is the GPU device?

easy

A. model = torch.nn.DataParallel(model, device_ids=[device])

B. model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[device])

C. model = torch.nn.DistributedDataParallel(model, device=device)

D. model = torch.nn.parallel.DistributedDataParallel(model, devices=device)

Solution

1. Step 1: Recall the correct syntax for DistributedDataParallel

   The correct class is torch.nn.parallel.DistributedDataParallel and it takes device_ids as a list of devices.

2. Step 2: Check each option for syntax correctness

   model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[device]) uses the correct class and parameter device_ids=[device]. model = torch.nn.DataParallel(model, device_ids=[device]) uses DataParallel, which is different. Options C and D use incorrect parameter names or class names.

3. Final Answer:

   model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[device]) -> Option B

4. Quick Check:

   Correct class and device_ids list = A [OK]

Hint: Use torch.nn.parallel.DistributedDataParallel with device_ids list [OK]

Common Mistakes:

• Using DataParallel instead of DistributedDataParallel
• Passing device instead of device_ids list
• Wrong class or parameter names

3. Consider the following code snippet for a distributed training setup:

import torch
import torch.distributed as dist
from torch.nn.parallel import DistributedDataParallel as DDP

def setup(rank, world_size):
    dist.init_process_group('gloo', rank=rank, world_size=world_size)

model = torch.nn.Linear(10, 1).to(rank)
model = DDP(model, device_ids=[rank])

print(model.module.weight.device)

What will be the output device of model.module.weight.device if rank=1 and GPUs are available?

medium

A. cuda:1

B. cpu

C. cuda:0

D. cuda:rank

Solution

1. Step 1: Understand device assignment in the code

   The model is moved to rank device with .to(rank), so for rank=1, it is on cuda:1.

2. Step 2: Check DistributedDataParallel wrapping

   DDP wraps the model and uses device_ids=[rank], so the model stays on cuda:1.

3. Final Answer:

   cuda:1 -> Option A

4. Quick Check:

   Model device matches rank GPU = B [OK]

Hint: Model device matches rank GPU in DDP setup [OK]

Common Mistakes:

• Assuming model stays on CPU
• Confusing device index with rank variable
• Thinking device is always cuda:0

4. You run a distributed training script using DistributedDataParallel but get the error: RuntimeError: Expected to have finished reduction in the prior iteration before starting a new one. What is the most likely cause?

medium

A. You forgot to call optimizer.step() after loss.backward()

B. You did not wrap the model with DistributedDataParallel

C. You used the wrong backend in init_process_group

D. You did not call dist.destroy_process_group() at the end

Solution

1. Step 1: Understand the error about unfinished reduction

   This error usually means gradients were not synchronized properly before the next iteration.

2. Step 2: Identify missing optimizer step

   Not calling optimizer.step() after loss.backward() can cause gradients to not be updated and reduction to remain unfinished.

3. Final Answer:

   You forgot to call optimizer.step() after loss.backward() -> Option A

4. Quick Check:

   Missing optimizer.step() causes reduction error = C [OK]

Hint: Always call optimizer.step() after backward() in DDP [OK]

Common Mistakes:

• Skipping optimizer.step() call
• Ignoring proper model wrapping
• Confusing backend errors with sync errors

5. You want to train a model on 4 GPUs using DistributedDataParallel across 2 machines, each with 2 GPUs. Which of the following is the correct approach to set up the training?

hard

A. Launch 4 processes on one machine only, ignoring the second machine

B. Launch 2 processes total, each process uses 2 GPUs, wrap model with DataParallel on each process

C. Launch 4 processes total, each process uses one GPU, initialize process group with correct rank and world_size, wrap model with DDP on each process

D. Launch 1 process per machine, use DDP with device_ids=[0,1] on each machine

Solution

1. Step 1: Understand multi-machine multi-GPU DDP setup

   Each GPU should have one process. For 4 GPUs across 2 machines, launch 4 processes total.

2. Step 2: Confirm process group and model wrapping

   Initialize the process group with correct rank and world_size=4. Wrap the model with DDP on each process using the assigned GPU.

3. Final Answer:

   Launch 4 processes total, each process uses one GPU, initialize process group with correct rank and world_size, wrap model with DDP on each process -> Option C

4. Quick Check:

   One process per GPU with correct init = D [OK]

Hint: One process per GPU, init process group correctly [OK]

Common Mistakes:

• Using fewer processes than GPUs
• Using DataParallel instead of DDP for multi-machine
• Ignoring second machine in setup