Enhance LLM runtime and installation script for Linux support

- Added a new function `pick_llamacpp_linux_asset_url` to select appropriate Linux release assets, prioritizing Ubuntu CUDA and Vulkan options while excluding Windows and macOS binaries.
- Updated the installation script to build `llama-server` from source when Linux CUDA binaries are unavailable, improving compatibility and flexibility.
- Revised documentation to reflect changes in asset handling and installation procedures.
- Added tests to validate the new asset selection logic, ensuring correct behavior in various scenarios.
This commit is contained in:
Leonid Pershin
2026-08-21 07:51:17 +03:00
parent 15b95f04c7
commit 242f5b7c89
4 changed files with 283 additions and 63 deletions
+3 -1
View File
@@ -120,6 +120,8 @@ Unit `gpu-rent-ollama` читает `/mnt/swarm_data/.gpu-rent-gpu.json`:
На `up`: скачать GGUF → `/mnt/swarm_data/llamacpp/models``llama-server` + systemd. Уже скачанные крупные файлы не трогает.
**Бинарник:** у upstream нет Linux CUDA в GitHub Releases (только Windows). `install_llamacpp.sh` собирает `llama-server` из исходников (`GGML_CUDA=ON`, arch из GPU probe). Если `nvcc` недоступен — fallback на Ubuntu Vulkan/CPU asset (без Windows). Pin: `LLAMACPP_TAG=b10545`. Override: `LLAMACPP_ASSET_URL` + `LLAMACPP_SHA256`.
### Пресеты (меню)
| # | ключ | что |
@@ -154,6 +156,6 @@ Busy (не гасить GPU):
```bash
OLLAMA_VERSION=0.6.5
OLLAMA_SHA256=<sha256 of ollama-linux-amd64.tgz>
LLAMACPP_TAG=b4690
LLAMACPP_TAG=b10545
# или LLAMACPP_ASSET_URL=... + LLAMACPP_SHA256=...
```