

In general I agree, but I’d put any conspiracy higher than AI, that’s just the scam de jour. It would be more along the lines of making conditions favourable for the death of personal computing and other boot on the face of humanity goals. It’s also likely this is a fairly natural evolution of late stage capitalism with technology, no conspiracy needed.
In this case it’s making lots and lots of GPU datacenters, which after a bubble pop will be useful for making compute a subscription for the masses and processing shittons of video feeds for the surveillance state, and the odd bit of AI.
This is why I don’t make much of the ‘the chips will be obsolete by the time they’re turned on’ argument, compute is compute, max efficiency or not. Obsolescence might sink a few (or many) companies, but the centralized compute assets will remain and need repurposing. Way too many of those purposes are harmful to the 99%.





Lemonade is pretty good, especially for AMD GPUs and the lazy, but it’s pretty opinionated (like ollama but not as toxic) and it’s not easy to directly control llama-server parameters. It feels a bit commercial / we want to IPO.
Seeing as you already have your parameters, llama-swap might be a better fit, lets you configure llama-server directly in a yaml, and switch between them. You can even have multiple versions of the same model (e.g. thinking or not, try out different parameters) and change without reloading.
Both work fine with e.g. Open WebUI and Pi (lemonade its own integration and llama-server needs an extension).