AmpereOne A192-32X 512GB (CPU only)
| GPU(s) | None (CPU only) |
|---|---|
| VRAM / unified memory (GB) | |
| CPU | AmpereOne A192-32X Amazon · eBay (used) · Newegg |
| Motherboard | |
| RAM | 512 GB Amazon · eBay (used) · Newegg |
| PSU | |
| Other | |
| Reported total cost | |
| Power | Peak 477 W (DeepSeek R1 671B) |
Measured speed (generation tokens/sec)
| Model | Quant | Engine | tok/s |
|---|---|---|---|
| DeepSeek R1 671B | llama.cpp | 4.18 | |
| Llama 3.1 70B | llama.cpp | 3.86 | |
| Llama 3.1 405B | llama.cpp | 0.9 | |
| Llama 3.2 3B | llama.cpp | 23.52 |
Numbers from Jeff Geerling's ai-benchmarks results table (llama.cpp/Ollama eval rate). Quantization not stated in the summary table. Power is peak whole-system draw during that test.
Source: https://github.com/geerlingguy/ai-benchmarks/issues/10
https://github.com/geerlingguy/ai-benchmarks
Blank fields = not stated in the source. We never estimate build data; GPU street prices above are separately sourced and dated.
Rent before you buy: try this model on a cloud GPU for a few dollars first: RunPod