[WIP]add npu support

[hipudding]

Add Ascend NPU as a new backend for pytorch-lighting

Fixes #19498

Before submitting

• Was this discussed/agreed via a GitHub issue? (not for typos and docs)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• Did you make sure to update the documentation with your changes? (if necessary)
• Did you write any new necessary tests? (not for typos and docs)
• Did you verify new and existing tests pass locally with your changes?
• Did you list all the breaking changes introduced by this pull request?
• Did you update the CHANGELOG? (not for typos, docs, test updates, or minor internal changes/refactors)

PR review

Anyone in the community is welcome to review the PR.
Before you start reviewing, make sure you have read the review guidelines. In short, see the following bullet-list:

Reviewer checklist

• Is this pull request ready for review? (if not, please submit in draft mode)
• Check that all items from Before submitting are resolved
• Make sure the title is self-explanatory and the description concisely explains the PR
• Add labels and milestones (and optionally projects) to the PR so it can be classified

---

📚 Documentation preview 📚: https://pytorch-lightning--19308.org.readthedocs.build/en/19308/

[gitguardian]

️✅ There are no secrets present in this pull request anymore.

If these secrets were true positive and are still valid, we highly recommend you to revoke them.
Once a secret has been leaked into a git repository, you should consider it compromised, even if it was deleted immediately.
Find here more information about risks.

---

<sup><sub>🦉 GitGuardian detects secrets in your source code to help developers and security teams secure the modern development process. You are seeing this because you or someone else with access to this repository has authorized GitGuardian to scan your pull request.

Our GitHub checks need improvements? Share your feedbacks!</sub></sup>

[hipudding]

Reference implementation #17700 #19443

[hhllxx1121]

Hello, if possible, could you show me a demo of how to use Lightning on an NPU?

[GuWei007]

@hipudding @hhllxx1121
More and more device manufacturers need to support their own backends. Can this be achieved by using the PrivateUse1 mechanism?

[hipudding]

Hello, if possible, could you show me a demo of how to use Lightning on an NPU?

Yes, it's very simple to use npu instead of other backends. Just replace accelerator to 'npu'. Note that this PR is under development and not pass all the tests. DO NOT use for production.

Here's official example with NPU.

# main.py
# ! pip install torchvision
import torch, torch.nn as nn, torch.utils.data as data, torchvision as tv, torch.nn.functional as F
import lightning as L

# --------------------------------
# Step 1: Define a LightningModule
# --------------------------------
# A LightningModule (nn.Module subclass) defines a full *system*
# (ie: an LLM, diffusion model, autoencoder, or simple image classifier).


class LitAutoEncoder(L.LightningModule):
    def __init__(self):
        super().__init__()
        self.encoder = nn.Sequential(nn.Linear(28 * 28, 128), nn.ReLU(), nn.Linear(128, 3))
        self.decoder = nn.Sequential(nn.Linear(3, 128), nn.ReLU(), nn.Linear(128, 28 * 28))

    def forward(self, x):
        # in lightning, forward defines the prediction/inference actions
        embedding = self.encoder(x)
        return embedding

    def training_step(self, batch, batch_idx):
        # training_step defines the train loop. It is independent of forward
        x, y = batch
        x = x.view(x.size(0), -1)
        z = self.encoder(x)
        x_hat = self.decoder(z)
        loss = F.mse_loss(x_hat, x)
        self.log("train_loss", loss)
        return loss

    def configure_optimizers(self):
        optimizer = torch.optim.Adam(self.parameters(), lr=1e-3)
        return optimizer


# -------------------
# Step 2: Define data
# -------------------
dataset = tv.datasets.MNIST(".", download=True, transform=tv.transforms.ToTensor())
train, val = data.random_split(dataset, [55000, 5000])

# -------------------
# Step 3: Train
# -------------------
autoencoder = LitAutoEncoder()
trainer = L.Trainer(accelerator='npu', devices='0,1', max_epochs=1, strategy='deepspeed')
trainer.fit(autoencoder, data.DataLoader(train), data.DataLoader(val))

[hipudding]

@hipudding @hhllxx1121 More and more device manufacturers need to support their own backends. Can this be achieved by using the PrivateUse1 mechanism?

As far as I know. pytorch-lightning use abstract accelerator, and not support plugins. If not all backends' accelerator can be merge. accelerator plugin is a good choice.