@MichaelTBacon true.
I was experimenting with ollama and lm studio in Jetbrains and VS Code, both for code generation and auto-completion.
Some or rather most models don't fit my GPU's RAM, so it's split in CPU and GPU and whenever I wrote code, the auto-completion thread would start working resulting in maximum fan rotations on my CPU cooler.
The qwen3-coder model is really good but doesn't fit in my GPU memory.
I tried smaller models, but they're a lot worse.
I run a bitcoin node on an old server xeon e3 1245v2. Every time there's a new transaction system load spikes.
So far, there have been no btc donations, so I'll probably shut it down again.
It requires over 1TB of HDD space, and when you're running a lightning node, even more