You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This document explains how to build and run the project with Docker, and how to enable GPU / NPU device passthrough.
4
+
5
+
1) Enable device passthrough (optional)
6
+
- If your system has an NPU, uncomment the following line in `docker/compose.yml` under `devices`:
7
+
- "/dev/accel:/dev/accel"
8
+
- If your system has a GPU, uncomment the following line in `docker/compose.yml` under `devices`:
9
+
- "/dev/dri:/dev/dri"
10
+
11
+
2) Build the image
12
+
Run from the project root or from `docker/`:
13
+
14
+
```
15
+
docker compose build
16
+
```
17
+
18
+
3) Export render group ID (for GPU access) and run
19
+
If you use GPU passthrough, export the `RENDER_GROUP_ID` environment variable so the container can access the render group:
20
+
21
+
```
22
+
export RENDER_GROUP_ID=$(getent group render | awk -F: '{printf "%s\n", $3}')
23
+
docker compose up -d
24
+
```
25
+
26
+
Notes
27
+
- If you don't have GPU/NPU hardware, you can leave the device lines commented; the app will still run but without hardware acceleration.
28
+
- If `getent group render` returns nothing, you may need to find the appropriate group ID for your system (often `render` or `video`) or run without setting `RENDER_GROUP_ID` and adjust permissions manually.
29
+
- The app is exposed on port `8080` by default; open http://localhost:8080 after the container is running.
0 commit comments