https://knowyourmeme.com/memes/loss https://knowyourmeme.com/memes/is-this-loss https://trending.knowyourmeme.com/editorials/guides/what-is-loss-the-classic-ctrlaltdel-webcomic-and-meme-explained

How well does it handle concurrency? I could see this as a vast improvement on llamacpp, but I was already speccing out vLLM.

A quick glance shows that I’m in the minority. I grabbed a Bosgames M5 with 128GB UMA RAM when I discovered openclaw (at the time claudebot). I’m running Lemonade, which is a wrapper for llama.cpp, sd-cpp, flux, and others. I’m happy to have paid a “mere” $2400 compared to what similar systems cost now, and now I’m toying with fine tuning my own models. Though, I primarily run Hermes Agent paired with Qwen3.6 35b A3B (using a heretic ablitieration for 0 refusals).

Hermes Agent does a lot of random stuff for me right now. It manages my calendar and to-do list in nextcloud via MCP server. It created and maintains an llm-wiki based on my Obsidian Vault. Then it creates a report every morning about both and the weather.

The server has tailscale installed, so yes I can access it anywhere.

Don’t forget that you contain an LLM that is constantly trying to pre-fill and decode at the same time on a constant agentic loop firing arbitrary cronjobs and skills. Stop hogging all the water and energy and polluting our water and air to produce nonsense that only other AIs can understand! (affectionate)

I have this exact setup (R730 and ZFS), but I’ll have to disagree on not using TrueNAS. There are features you may want to use, and the logical separation of the zfspool from the rest of the server has been handy. I boot and store my VMs off of SSDs outside of the main NAS pool.

If you want to use a NVMe boot drive on a PCIE card, the server isn’t natively capable of it. You need to use a USB drive to bootstrap it with Clover. I forget the exact technical details. I have had no problems leaving it in the internal usb port over a couple years so far.