te0006

Read TFA. They were very aware of the dangers of such an approach. So they avoided it to the extent possible.

I was there three days ago, can comfirm. And if you want to stay overnight, you _of course_ sleep in the neighbouring IKEA Hotel! And if you check-in to the hotel first, you‘ll get a voucher for free entrance to the museum. Did I mention that an absolutely excellent breakfast is included in the very reasonable room rate? And in case you visit the museum in the near future, they currently have a special (temporary) exhibition about IKEA's founder, Ingvar Kamprad, who was a very creative but also exceptionally socially and environmentally responsible entrepreneur. Nice contrast to today's appalling tech billionaires.

Interesting setup. What GPU(s)/VRAM, CPU and RAM are you using for the 122B model, with which quantization, and what token rates do you achieve for prefill and generation?

"it says" - where did you find info on HW support? Or did you try to build and flash the thing yourself?

Yes, same for the case where the technology would have been found embodied in machinery aboard a crashed UFO.

Does anybody think Aluminium as a brand name is a good choice? Especially considering the intended expansion towards the premium market. To me it sounds cheap, second-rate, ersatz. What you use if you cannot afford a better metal. Chrome is shiny, aluminium surfaces soon get dim again after any polishing attempt.

If you have a south (or SE or SW) facing wall without much shadowing from nearby buildings or trees, vertikal mounting does work OK. Do not expect to reach the panel's nominal Wp rating though, output will peak at 50-70% of that. But panels are cheap - if there's enough room, just overprovision twofold. Just take care to buy an inverter that is OK with such a bigger configuration. And vertically mounted panels will generate more power off-season than tilted ones.

Note that the small print on the GMKtec site says that prices do not include customs and VAT. Which seem to amount to 19% in the EU. So, almost 2,4K€.

Interesting - do you need to take any special measures to get OSS genAI models to work on this architecture? Can you use inference engines like Ollama and vLLM off-the-shelf (as Docker containers) there, with just the Radeon 8060S GPU? What token rates do you achieve?

(edit: corrected mistake w.r.t. the system's GPU)