Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
132.7 tok/s
TTFT
1459 ms
Safe context
53K
Memory
38.5 GB / 128.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 132.7 tok/s | 2122 ms | 94K |
| Chat | C | Runs well | 132.7 tok/s | 796 ms | 28K |
| Coding | C | Runs well | 132.7 tok/s | 1459 ms | 53K |
| RAG | C | Runs well | 132.7 tok/s | 2653 ms | 94K |
| Reasoning | C | Runs well | 136.5 tok/s | 1676 ms | 53K |
How Qwen 2.5 Coder 32B (32B params) fits at each quantization level on Gaudi 3 128GB (128.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 12.5 GB | Low | D32 |
Q3_K_S | 3 | 15.7 GB | Low | D32 |
NVFP4 | 4 |
ollama run qwen-2.5-coder-32bhuggingface-cli download qwen-2.5-coder-32b17.9 GB |
| Medium |
| D32 |
Q4_K_M | 4 | 19.5 GB | Medium | D33 |
Q5_K_M | 5 | 23.0 GB | High | D33 |
Q6_K | 6 | 26.2 GB | High | D34 |
Q8_0 | 8 | 34.2 GB | Very High | D35 |
F16Best for your GPU | 16 | 65.6 GB | Maximum | C40 |