Written by Jakub Rusinowski · Last updated February 28, 2024
Yes, but it is tight
Yes, but it is tight — StarCoder 2 15B at Q5_K_M needs about 12.5 GB of the 12.8 GB usable on 16 GB system RAM, leaving only ~0.3 GB before the runtime starts swapping. Expect ~5.9 tok/s (estimated), with room for about 8,192 tokens of context.
Confidence: medium · Recommended quantization: Q5_K_M · Estimated speed: ~5.9 tok/s
| Usable memory for models | 12.8 GB |
| Memory bandwidth | 90 GB/s |
| Quant | Memory needed | Fits 12.8 GB? | Max context | Est. speed | Download |
|---|---|---|---|---|---|
| F16 | 32.5 GB | ✗ No | — | — | 31 GB |
| Q8_0 | 17.9 GB | ✗ No | — | — | 16.5 GB |
| Q6_K | 14.2 GB | ✗ No | — | — | 12.7 GB |
| Q5_K_M | 12.5 GB | ✓ Yes | 8K | ~5.9 tok/s | 11 GB |
| Q4_K_M | 10.8 GB | ✓ Yes | 16K | ~6.8 tok/s | 9.4 GB |
| Q3_K_M | 8.1 GB | ✓ Yes | 16K | ~9.5 tok/s | 6.6 GB |
| Q2_K | 6.6 GB | ✓ Yes | 16K | ~12 tok/s | 5.1 GB |
| Variant | Needs at Q4_K_M | Fits? | Est. speed |
|---|---|---|---|
| StarCoder 2 15B | 10.8 GB | ✓ Fits | ~6.8 tok/s |
| StarCoder 2 7B | 5.7 GB | ✓ Fits | ~14.1 tok/s |
| StarCoder 2 3B | 2.9 GB | ✓ Fits | ~32.2 tok/s |
llama.cpp (CPU build) or Ollama — both run without a GPU
Yes, but it is tight — StarCoder 2 15B at Q5_K_M needs about 12.5 GB of the 12.8 GB usable on 16 GB system RAM, leaving only ~0.3 GB before the runtime starts swapping. Expect ~5.9 tok/s (estimated), with room for about 8,192 tokens of context.
Q5_K_M — it needs about 12.5 GB of the 12.8 GB available, downloads as roughly 11 GB, and runs at an estimated 5.9 tokens/sec with up to 8K of context.
Nothing binding — the model fits with headroom and generates at a usable speed on this hardware.
llama.cpp (CPU build) or Ollama — both run without a GPU
← Can I Run It? | StarCoder 2 model page | Check your hardware