> but VMs are pretty easy so just went that route. Don’t you need at least 2 GPU... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		nicce on Jan 31, 2025 \| parent \| context \| favorite \| on: Llama.cpp supports Vulkan. why doesn't Ollama? > but VMs are pretty easy so just went that route. Don’t you need at least 2 GPUs in that case and put kernel level passthrough?

a12k on Jan 31, 2025 | [–]

I don’t use GPU. Works fine, but the large Mixtral models are slow.

bdhcuidbebe on Jan 31, 2025 | [–]

i pass through my dGPU to VM and use iGPU for desktop

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact