# GPU vs CPU

NeoHive runs locally and uses your hardware to index and search your code and knowledge. This runs faster with a GPU but works fine on CPU.

## Auto-detection

The installer automatically detects your hardware and selects the best backend:

* **NVIDIA GPU** : uses CUDA
* **AMD GPU** : uses ROCm
* **Intel/other GPU** : uses Vulkan
* **No GPU** : falls back to CPU

**macOS (Apple Silicon):** Macs with M-series chips run the CPU backend. NeoHive is optimised for Apple Silicon, so performance is excellent : no GPU backend needed. The same applies to ARM Linux hosts like AWS Graviton.

In most cases, you don't need to configure anything.

## Forcing CPU mode

If GPU detection picks the wrong backend, or the GPU backend fails to start, force CPU mode:

```sh
NEOHIVE_BACKEND=cpu bash <(curl -fsSL https://raw.githubusercontent.com/NeoHiveAI/install/main/install.sh)
```

CPU mode runs everywhere but is slower for indexing large repositories. Day-to-day use is fast on either backend.

If the installer detects a backend issue, it will suggest this command automatically.

## NVIDIA Container Toolkit

If you have an NVIDIA GPU and the container fails to start, make sure the [NVIDIA Container Toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html) is installed. The installer will tell you if it's missing.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.neohive.ai/configuration/gpu.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.