noonghunna/club-3090
Community recipes for serving LLMs on RTX 3090/4090/5090 CUDA gpus. Multi-engine (vLLM, sglang, llama.cpp and other custom engines) and model-agnostic. Currently shipping Qwen3.8-27B Qwen3.6 35B Gemma 4 26B Gemma 4 31B configs for 1×/2×/4x/8x cards.
Build with Backblaze B2
SDKs, agent skills, IDE extensions, and reference pipelines from Backblaze Labs. All open source.
Loading star history...
Where it stands
Rank, growth pace and release cadence relative to everything GStars tracks.
2 releases in the last 90 days, the latest 31 days ago, with 33 contributors on GitHub.
Use Cases & Benefits
- Provides community-curated recipes and configurations for serving large language models on RTX 3090/4090/5090 GPUs using multiple inference engines.
- Enables optimized, model-agnostic local LLM serving with validated Docker compose setups and hardware-aware tuning for maximum throughput and robustness.
- Use for running modern LLMs at home or in a homelab with 1 or 2 RTX 3090 GPUs for development or experimentation.
- Use for benchmarking and comparing multi-engine LLM inference performance and stability across different GPU setups and models.
- Use for deploying multi-model AI workloads including vision and tool-augmented agents with seamless switching and health monitoring via a terminal UI cockpit.
Should you depend on it?
License terms, project hygiene and risks, from GitHub metadata and GStars' snapshots.
License
Apache 2.0
Permissive: commercial use, modification and redistribution are allowed with attribution; includes an explicit patent grant from contributors.
Community files
- README: present
- Contributing guide: present
Watch-outs
No watch-outs from GStars' checks.
General guidance from repository metadata, not legal advice. How we score
From the README
Recipes for serving LLMs locally on RTX 3090s. Multi-engine (vLLM, llama.cpp, ik_llama), multi-model, model-agnostic by design.
If you have one or two RTX 3090s and want to run modern LLMs at home, in a homelab, or as a dev backend — this repo collects the working configs, patches, and benchmarks.
🎯 4090 or 5090 owner? The composes run cross-rig — contributors have benched both with measured numbers: Can I use a 4090? → · Can I use a 5090? →. The tooling is calibrated for 3090s but the configs are class-aware; per-class gotchas (4090's tighter idle VRAM, 5090's 32 GB envelope) + cross-rig benchmark rows live in the FAQ.
🎨 Want image generation too? The Image Studio bundle runs Ideogram-4 image gen + a chat model + Open WebUI together on two GPUs — one command:
bash scripts/setup-image-studio.sh.🪟 On Windows? These steps assume Linux/macOS. Set up WSL2 first → docs/WSL_SETUP.md (start-to-finish). Native Windows runs only the upstream llama.cpp binary — none of this repo's tooling.
launch.shcallsswitch.sh(down old, up new) and thenverify-full.shso you know it's serving cleanly before you point a client at it. Seescripts/for all helpers.
Daily snapshots
Stars gained per day, and how forks and open issues moved, from GStars' daily snapshots
+317 stars over 60 days across 6 snapshot days; best day +11 on Sep 2.
Forks
169+38
Open issues
87+57
Compared with peers in Backend
Fastest-growing repositories GStars tracks in this cohort for the week of Nov 1, by 7-day star growth
| Repository | Stars | +7d | Health | Last release |
|---|---|---|---|---|
| anthropics/skills | 187.8K | +2.9K | 15 | — |
| DuarteSantos8/openGym | 19.7K | +2.4K | 60 | 2 days ago |
| shiyu-coder/Kronos | 47.0K | +2.3K | 16 | — |
| NationalSecurityAgency/ghidra | 98.2K | +2.3K | 64 | 14 days ago |
| microsoft/VibeVoice | 56.0K | +1.1K | 29 | — |
| noonghunna/club-3090 | 2.5K | — | 58 | 31 days ago |
| Category median | 56.0K | +2.3K | 29 | — |
Discover Repositories
Search across tracked repositories by name or description