Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
60.4 tok/s
TTFT
3204 ms
Safe context
32K
Memory
32.1 GB / 64.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 57.1 tok/s | 4935 ms | 55K |
| Chat | C | Runs well | 57.1 tok/s | 1851 ms | 17K |
| Coding | C | Runs well | 60.4 tok/s | 3204 ms | 32K |
| RAG | C | Runs well | 57.1 tok/s | 6169 ms | 55K |
| Reasoning | C | Runs well | 57.1 tok/s | 4010 ms | 32K |
How Qwen 2.5 Coder 32B (32B params) fits at each quantization level on AMD Instinct MI210 64GB (64.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 12.5 GB | Low | D34 |
Q3_K_S | 3 | 15.7 GB | Low | D35 |
NVFP4 | 4 |
ollama run qwen-2.5-coder-32bhuggingface-cli download qwen-2.5-coder-32b17.9 GB |
| Medium |
| D35 |
Q4_K_M | 4 | 19.5 GB | Medium | D36 |
Q5_K_M | 5 | 23.0 GB | High | D37 |
Q6_K | 6 | 26.2 GB | High | D38 |
Q8_0Best for your GPU | 8 | 34.2 GB | Very High | C41 |
F16 | 16 | 65.6 GB | Maximum | F0 |