Step 3: Choose Your GPU Platform¶
Before downloading templates, determine which GPU acceleration you want to use. This affects which server template (for video transcoding) and which machine learning template (for face recognition and image search) you need.
graph TD
A[What GPU do you have?] --> B{NVIDIA?}
A --> C{Intel iGPU?}
A --> D{AMD?}
A --> E{No GPU / CPU only}
B --> B1[Server: immich-server-nvenc]
B --> B2[ML: immich-machine-learning-cuda]
C --> C1[Server: immich-server-qsv-vaapi]
C --> C2[ML: immich-machine-learning-openvino]
D --> D1[Server: immich-server-qsv-vaapi]
D --> D2[ML: immich-machine-learning-rocm]
E --> E1[Server: immich-server]
E --> E2[ML: immich-machine-learning]
| GPU | Server Template (Transcoding) | ML Template (Inference) |
|---|---|---|
| None / CPU only | immich-server |
immich-machine-learning |
| Intel iGPU (N100, UHD, Iris) | immich-server-qsv-vaapi |
immich-machine-learning-openvino |
| AMD (Polaris+) | immich-server-qsv-vaapi |
immich-machine-learning-rocm |
| NVIDIA (Pascal+) | immich-server-nvenc |
immich-machine-learning-cuda |