# Device Type

## CHI\@Edge Supported Device Types

This page details the Single Board Computers (SBCs) available for edge computing experiments. These devices serve as the primary compute nodes and hosts for peripherals.

| machine\_name (for reservation) | Description                                  |
| ------------------------------- | -------------------------------------------- |
| raspberrypi5                    | Raspberry Pi 5, all variants                 |
| raspberrypi4-64                 | Raspberry Pi 4, all variants, including CM4  |
| jetson-nano                     | Nvidia Jetson Nano                           |
| jetson-xavier-nx-devkit-emmc    | Nvidia Jetson Xavier NX                      |
| jetson-orin-nano-devkit-nvme    | Nvidia Jetson Orin Nano Developer Kit (NVMe) |
| jetson-agx-orin-devkit-64gb     | Nvidia Jetson AGX Orin 64GB Developer Kit    |

## Hardware Details

### Raspberry Pi 5

<div align="left"><figure><img src="/files/QQCO5WLt4QKe1HheQfG3" alt="" width="375"><figcaption></figcaption></figure></div>

* 8gb and 16gb variants
* <https://www.raspberrypi.com/products/raspberry-pi-5/>

#### Compute Module 5 (coming soon)

* <https://www.raspberrypi.com/products/compute-module-5/?variant=cm5-104032>
* Lower power draw, wide temperature range support, suitable for embedding into outdoor or rugged applications.

### Raspberry Pi 4

#### Model B, 8gb

* <https://www.raspberrypi.com/products/raspberry-pi-4-model-b/specifications/>

#### compute module 4, 8gb

<div align="left"><figure><img src="/files/rHpwxPEFwRDspYkHPcBZ" alt="" width="300"><figcaption></figcaption></figure></div>

* <https://www.raspberrypi.com/products/compute-module-4/?variant=raspberry-pi-cm4001000>
* Lower power draw, wide temperature range support, suitable for embedding into outdoor or rugged applications.

### Nvidia Jetson Nano

<div align="left"><figure><img src="/files/zWAFHFA1rRYQnxQjPy2G" alt="" width="375"><figcaption></figcaption></figure></div>

* <https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-nano/product-development/>

The latest Nvidia software supported on the Jetson Nano is:

* [JetPack 4.6.6](https://developer.nvidia.com/jetpack-sdk-466)
* [L4T 32.7.6](https://developer.nvidia.com/embedded/linux-tegra-r3276)

This is based on Ubuntu 18.04, Linux Kernel 4.9, and [**CUDA 10.2**](https://docs.nvidia.com/cuda/archive/10.2/cuda-toolkit-release-notes/index.html#title-new-features)

**You must use a container image targeting CUDA10.2 to work on these devices.**

The NVIDIA SDK for these devices is unfortunately end-of-life, and we are rolling out support for the replacement "Orin Nano" devices, described below.

### Nvidia Jetson Xavier NX

<div align="left"><figure><img src="/files/JaAnf9JUkRk0akFM7sYZ" alt=""><figcaption></figcaption></figure></div>

* <https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-xavier-series/>

The latest Nvidia software supported on the Xavier NX is:

* [JetPack 4.6.6](https://developer.nvidia.com/jetpack-sdk-466)
* [L4T 32.7.6](https://developer.nvidia.com/embedded/linux-tegra-r3276)

This is based on Ubuntu 18.04, Linux Kernel 4.9, and [**CUDA 10.2**](https://docs.nvidia.com/cuda/archive/10.2/cuda-toolkit-release-notes/index.html#title-new-features).

**You must use a container image targeting CUDA 10.2 to work on these devices.**

Support for these is limited, as NVIDIA considers them EoL. The replacement is the AGX Orin and Orin Nano, described below.

### Jetson Orin Nano Developer Kit

<div align="left"><figure><img src="/files/Xyt33PnWkQQ159UF4ySx" alt="" width="563"><figcaption></figcaption></figure></div>

* <https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-orin/>
* <https://nvdam.widen.net/s/zkfqjmtds2/jetson-orin-datasheet-nano-developer-kit-3575392-r2>

The Orin Nano is Nvidia's current offering, replacing the Jetson Nano. CHI\@Edge images target the NVMe variant of the developer kit (`jetson-orin-nano-devkit-nvme`);

The Nvidia software supported on the Orin Nano is:

* JetPack 6.x
* L4T r36.4
* **CUDA 12.6**

This is based on Ubuntu 22.04. The Orin Nano is currently running in the original (non-Super) power profile.

Container images must target **CUDA 12.6** (or compatible) and `linux/arm64`.

### Jetson AGX Orin 64GB Developer Kit

<div align="left"><figure><img src="/files/StCPHY706T2MLbe7JMyC" alt="" width="375"><figcaption></figcaption></figure></div>

* <https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/jetson-orin/>
* [Jetson AGX Orin Developer Kit datasheet](https://developer.nvidia.com/embedded/downloads)

The AGX Orin is the high-end Orin-series board. CHI\@Edge currently supports the 64GB developer kit (`jetson-agx-orin-devkit-64gb`).

The Nvidia software supported on the AGX Orin is:

* JetPack 6.x
* L4T r36.4
* **CUDA 12.6**

This is based on Ubuntu 22.04.

Container images must target **CUDA 12.6** (or compatible) and `linux/arm64`.

## Hardware Summary Table

<table data-full-width="true"><thead><tr><th>Device Name</th><th width="117.265625">CPU</th><th>Memory</th><th>GPU Arch</th><th>AI Perf [1]</th><th>Power (TDP)</th></tr></thead><tbody><tr><td><strong>Raspberry Pi 4B</strong></td><td>4c A72<br>@ 1.8GHz</td><td>8GB LPDDR4<br>13.3 GB/s</td><td>VideoCore VI</td><td></td><td>3W - 7W</td></tr><tr><td><strong>Raspberry Pi 5</strong></td><td>4c A76<br>@ 2.4GHz</td><td>8GB LPDDR4x<br>27.3 GB/s</td><td>VideoCore VII</td><td></td><td>5W - 12W</td></tr><tr><td><strong>Jetson Nano</strong></td><td>4c A57<br>@ 1.43GHz</td><td>4GB LPDDR4<br>25.6 GB/s</td><td>Maxwell (128c)</td><td>472 GFLOPS</td><td>5W - 10W</td></tr><tr><td><strong>Jetson Xavier NX</strong></td><td>6c Carmel v8.2</td><td>8GB LPDDR4x<br>59.7 GB/s</td><td>Volta (384c+48T)</td><td>21 TOPS (Dense, Int8)</td><td>10W - 20W</td></tr><tr><td><strong>Jetson Orin Nano (Dev Kit)</strong></td><td>6c A78AE<br>@ 1.5GHz</td><td>8GB LPDDR5<br>68 GB/s</td><td>Ampere (1024c + 32T)</td><td>40 TOPS (Sparse, Int8)</td><td>7W - 15W</td></tr><tr><td><strong>Jetson AGX Orin 64GB</strong></td><td>12c A78AE<br>@ 2.2GHz</td><td>64GB LPDDR5<br>204.8 GB/s</td><td>Ampere (2048c + 64T)</td><td>275 TOPS (Sparse, Int8)</td><td>15W - 60W</td></tr></tbody></table>


---

# 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://chameleoncloud.gitbook.io/chi-edge/hardware-info/device-type.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.