Tangentially related but what does a “Radeon 8060S” mean? I’m so confused what the memory bandwidth is and how much VRAM (I think it’s shared?). I was gonna buy a M2 studio for local LLM but this seems like a good choice. But I can’t figure out what the token/s would be like. I have Qwen2.5 3B running on a Mac mini M1 and it’s not great. Slow and gets stuff wrong. If I can get the 14B running with a response time below 3 seconds for Home Assistant, it would be perfect.

Edit: found it, 256 GB/s for the shared memory bandwidth