Framework Desktop (AMD Strix Halo) mainboard
| GPU(s) | AMD Ryzen AI Max+ 395 (Radeon 8060S iGPU; Ollama runs CPU only, llama.cpp runs on iGPU via Vulkan) Amazon · eBay (used) · Newegg |
|---|---|
| VRAM / unified memory (GB) | |
| CPU | AMD Ryzen AI Max (Strix Halo) Amazon · eBay (used) · Newegg |
| Motherboard | Framework Desktop mainboard Amazon · eBay (used) · Newegg |
| RAM | |
| PSU | |
| Other | Ollama benchmark |
| Reported total cost | |
| Power | ~141 W during Ollama CPU runs; ~133 W during llama.cpp Vulkan iGPU runs (Llama 3.2 3B, Llama 3.1 70B) |
Measured speed (generation tokens/sec)
📏 How measured: reported by the original source; RigBase re-checked the numbers at that source
| Model | Quant | Engine | tok/s | Details |
|---|---|---|---|---|
| DeepSeek R1 Distill 8B | Ollama default | Ollama (CPU only, iGPU not used) | 19.49 | |
| DeepSeek R1 Distill 70B | Ollama default | Ollama (CPU only, iGPU not used) | 2.43 | |
| Llama 3.1 70B | Q4_K_M | llama.cpp Vulkan (iGPU, llama-bench tg128) | 4.97 | |
| Llama 3.2 3B | Q4_K_M | llama.cpp Vulkan (iGPU, llama-bench tg128) | 88.14 |
Source: https://github.com/geerlingguy/ai-benchmarks/issues/21
https://github.com/geerlingguy/ai-benchmarks/issues/21#issuecomment-3164568218
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 (referral link; sign up with Google to get the new-user bonus)