> ## Documentation Index
> Fetch the complete documentation index at: https://docs.veri.studio/llms.txt
> Use this file to discover all available pages before exploring further.

# Hot Aisle (AMD MI300X)

> Serve on AMD MI300X: 192 GB of HBM3 on a single GPU — 70B-class models in bf16 without tensor parallel, on the ROCm vLLM stack.

Hot Aisle serves on AMD MI300X VMs. The headline: **192 GB of HBM3 on a single GPU** — enough to serve a 70B model in bf16 on one card, where NVIDIA equivalents need a multi-GPU node.

## Serving shapes

| `gpu_type`     | Memory | GPUs per node | Status    |
| -------------- | ------ | ------------- | --------- |
| `MI300X-192GB` | 192 GB | 1             | Available |

Serving is single-GPU per deployment for now (multi-GPU ROCm tensor parallel is unvalidated).

## Creating a deployment

```python theme={null}
client.deployments.create(
    model="Qwen/Qwen2.5-7B-Instruct",
    source="huggingface",
    name="my-amd-deploy",
    gpu={"gpu_type": "MI300X-192GB", "gpu_count": 1},
    provider="hotaisle",
)
```

## Engines

The default `vllm` engine runs the ROCm vLLM container.

<Warning>
  `engine="max"` is not yet supported on AMD GPUs — a create returns a 400 naming the vllm alternative. (MAX on MI300X currently crash-loops after passing health; the gate lifts once the MAX ROCm path stabilizes.)
</Warning>

## Notes

* Boxes run ROCm, not CUDA; the serving stack is the ROCm vLLM docker image, so the OpenAI-compatible surface is identical to NVIDIA deployments.
* Check live MI300X capacity before creating: `veri gpu list --provider hotaisle` shows a real-time quantity.
* Training on the same silicon: see [Hot Aisle for training](/training/providers/hotaisle), including MI300X performance tuning.

## Where to go next

<CardGroup cols={2}>
  <Card title="All serving providers" icon="server" href="/deployments/providers">
    Compare shapes, status, and the live CLI catalog.
  </Card>

  <Card title="Train on MI300X" icon="graduation-cap" href="/training/providers/hotaisle">
    The training side of the AMD stack.
  </Card>
</CardGroup>