jazzymc/infrastructure
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs(services): add Ollama iGPU (Radeon 780M / ROCm) entry
ci/woodpecker/push/woodpecker Pipeline was successful
Details

Browse Source 
- Document ollama migrated to Dockge stack on ollama/ollama:rocm with
  gfx1103 enablement (HSA_OVERRIDE_GFX_VERSION=11.0.0, OLLAMA_IGPU_ENABLE=1)
- Verified Ornith 1.0 9B runs 100% on the 780M iGPU (~11 tok/s)
- CHANGELOG entry for 2026-06-29
- (also includes previously-uncommitted Development services section
  already present in the working tree)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Kaloyan Danchev
2026-06-29 21:30:43 +03:00
parent 08874ca0b3
commit 3a28e42543
2 changed files with 74 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/03-SERVICES-OTHER.md
+66 -13
View File
@@ -1,6 +1,6 @@
# Other Services
**Last Updated:** 2026-05-05
**Last Updated:** 2026-06-29
Non-critical services that enhance functionality but don't affect core network operation.
@@ -308,18 +308,6 @@ Non-critical services that enhance functionality but don't affect core network o
**Purpose:** Mesh VPN for remote access
---
## AI / LLM
### Ollama
| Component | Port |
|-----------|------|
| ollama | 11434 |
**Purpose:** Local LLM inference server
### Open WebUI
| Component | Port |
@@ -415,6 +403,42 @@ Non-critical services that enhance functionality but don't affect core network o
---
## Development
### OpenVSCode Server (Web IDE)
| Network | Port | URL |
|---------|------|-----|
| host | 3100 | https://code.xtrm-lab.org |
**Purpose:** Web-based VS Code IDE with full host access
**Auth:** Authentik forward auth (SSO)
**Binary:** `/mnt/user/appdata/openvscode/current/` (host-native, not a container)
**Config:** `/mnt/user/appdata/openvscode/config/`
**Workspace:** `/mnt/user/projects/` (12 personal repos + 18 AMPECO projects)
**See:** `12-DEVELOPMENT-ENVIRONMENT.md`
### Claude Code
| Network | Version |
|---------|---------|
| host | 2.1.71 |
**Purpose:** AI coding assistant CLI
**Binary:** `/mnt/user/appdata/claude-code/.npm-global/bin/claude`
**Config:** `/mnt/user/appdata/claude-code/.claude/`
### Cooperator CLI
| Network | Version |
|---------|---------|
| host | 3.36.1 |
**Purpose:** AI-driven development workflow automation (AMPECO)
**Binary:** `/usr/local/lib/node_modules/@ampeco/cooperator/`
---
## Container Management
### Dockge
@@ -466,6 +490,35 @@ Non-critical services that enhance functionality but don't affect core network o
---
## AI / LLM
### Ollama (Local LLM Inference)
| Network | Port |
|---------|------|
| bridge | 11434 |
**Purpose:** Local LLM inference, GPU-accelerated on the integrated AMD Radeon 780M (gfx1103) via ROCm 7.2.
**Image:** `ollama/ollama:rocm` (Dockge stack — replaced old `dockerman` template `my-ollama.xml`)
**API:** http://192.168.10.20:11434
**Data:** `/mnt/user/appdata/ollama`
**Stack:** `/mnt/user/appdata/dockge/stacks/ollama/compose.yaml`
**iGPU enablement** (gfx1103 not on AMD's official ROCm allowlist):
| Setting | Value |
|---------|-------|
| `HSA_OVERRIDE_GFX_VERSION` | 11.0.0 (report 780M as gfx1100) |
| `OLLAMA_IGPU_ENABLE` | 1 (Ollama 0.30+ drops iGPUs otherwise) |
| `OLLAMA_FLASH_ATTENTION` | 1 |
| `OLLAMA_KV_CACHE_TYPE` | q8_0 |
| devices | `/dev/kfd`, `/dev/dri` |
| group_add | `18` (video) |
**Verified:** Ornith 1.0 9B GGUF runs 100% on GPU (33/33 layers), ~11 tok/s. iGPU is RAM-bandwidth bound (shared GTT, ~23 GiB addressable). 9B fits; 35B RAM-tight (no swap); 397B not feasible.
---
## Stopped/Disabled Services
| Service | Reason | Status |