Written by Jakub Rusinowski · Last updated February 28, 2024
Yes — comfortably
Yes, comfortably — StarCoder 2 15B at Q8_0 needs about 17.9 GB of the 32 GB usable on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM), leaving ~14.1 GB spare and running at ~67.8 tok/s (estimated), with room for about 16,384 tokens of context.
Confidence: high · Recommended quantization: Q8_0 · Estimated speed: ~67.8 tok/s
| Usable memory for models | 32 GB |
| Memory bandwidth | 1792 GB/s |
| Form factor | Desktop |
| Operating system | Windows or Linux |
| Memory upgradeable | Yes |
| Quant | Memory needed | Fits 32 GB? | Max context | Est. speed | Download |
|---|---|---|---|---|---|
| F16 | 32.5 GB | ✗ No | — | — | 31 GB |
| Q8_0 | 17.9 GB | ✓ Yes | 16K | ~67.8 tok/s | 16.5 GB |
| Q6_K | 14.2 GB | ✓ Yes | 16K | ~83.7 tok/s | 12.7 GB |
| Q5_K_M | 12.5 GB | ✓ Yes | 16K | ~93.7 tok/s | 11 GB |
| Q4_K_M | 10.8 GB | ✓ Yes | 16K | ~105.7 tok/s | 9.4 GB |
| Q3_K_M | 8.1 GB | ✓ Yes | 16K | ~134.9 tok/s | 6.6 GB |
| Q2_K | 6.6 GB | ✓ Yes | 16K | ~159.1 tok/s | 5.1 GB |
| Variant | Needs at Q4_K_M | Fits? | Est. speed |
|---|---|---|---|
| StarCoder 2 15B | 10.8 GB | ✓ Fits | ~105.7 tok/s |
| StarCoder 2 7B | 5.7 GB | ✓ Fits | ~176.1 tok/s |
| StarCoder 2 3B | 2.9 GB | ✓ Fits | ~271.2 tok/s |
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
Yes, comfortably — StarCoder 2 15B at Q8_0 needs about 17.9 GB of the 32 GB usable on RTX 5090 Desktop (32 GB VRAM, 64 GB RAM), leaving ~14.1 GB spare and running at ~67.8 tok/s (estimated), with room for about 16,384 tokens of context.
Q8_0 — it needs about 17.9 GB of the 32 GB available, downloads as roughly 16.5 GB, and runs at an estimated 67.8 tokens/sec with up to 16K of context.
Nothing binding — the model fits with headroom and generates at a usable speed on this hardware.
Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving
← Can I Run It? | StarCoder 2 model page | Check your hardware