This has got to be an LLM. The post history is just more of the same. But I don’t get it, why would anyone use an LLM for lemmy responses?

I’ve had HA hang while running wyoming-openwakeword, but my setup is pretty different from yours – I was trying to get wakeword detection off the audio track in an RTSP stream. I didn’t solve it, and gave up.

Wendell and Serve The Home in general have totally lost the “home” focus. Sure, I love reading about cutting edge 800GbE transceivers… but they hardly ever cover actually practical home stuff anymore. Their network switch reviews are the only thing vaguely applicable left.

I run a 1600X as a Proxmox host, I have used TLP to set all of the configurable that are exposed, and it idles at 70W. There are a few other things consuming power though - a spinning HDD, a discrete GPU (quadro K2200) doing zero work, and four DIMMs.

This is helpful to heat my apartment in the winter, but regrettable overall.