Hacker News: AbsurdCensor

New comment by AbsurdCensor in "Ubiquiti: Enterprise NAS, Built on ZFS"

AbsurdCensor — Thu, 18 Jun 2026 16:06:53 +0000

I started with Unifi and it's been pretty great overall. I've integrated all the cameras into Home Assistant, it's all local, and can bridge with HomeBridge so it all shows up and plays nicely with HomeKit as well. Rock solid and very few complaints.

New comment by AbsurdCensor in "Ubiquiti: Enterprise NAS, Built on ZFS"

AbsurdCensor — Thu, 18 Jun 2026 16:04:55 +0000

That's without storage. They are charging $750 each for 24tb HDD's, so filling it up brings that cost to $16k. Only need to run it for 13+ years and have zero HDD failures in that time, and then pay for all the media you are going to load it up with. Not exactly sure this would be cheaper or easier than just paying for streaming services and cancelling them when you don't need them.

New comment by AbsurdCensor in "Midjourney Medical"

AbsurdCensor — Thu, 18 Jun 2026 12:17:17 +0000

I thought this was the main goal of things like preventative medicine. Earlier testing, and more frequent testing allows you to catch things while you still 'feel healthy' so they do not become a chronic problem in the long term. Simple things like tracking weight, blood pressure over time. Add in things like colonoscopies, breast exams, I would say most medicine should be preventative, ie keeping you healthy, rather than reactive, of just trying to try you when you are ill.

New comment by AbsurdCensor in "Running local models is good now"

AbsurdCensor — Tue, 16 Jun 2026 18:08:42 +0000

I've certainly thought about just moving the box to Linux, but it took far to long personally to get everything running under AMD and it works 'well enough' that I don't want to make the switch. I tried playing with GAIA on it, felt a bit limited, and now have Hermes up and running, and that seems to work quite well. All the tools are changing so quickly, it's sometimes difficult to settle in on 'what's best', so I certainly can understand folks that just want to pay for a AI subscription and be done with it.

New comment by AbsurdCensor in "Running local models is good now"

AbsurdCensor — Tue, 16 Jun 2026 15:48:00 +0000

I think currently you can only get the M3 Ultra Studio with 96gb, and for coding tasks, say you rub Qwen Coder on it (which doesn't need that much ram), it's not the fastest, something like 30-40 tok/sec. Probably better with a MacBook Pro with the M5 chip. There is a website for comparing different configurations and models: https://llmcheck.net/benchmarks

New comment by AbsurdCensor in "Running local models is good now"

AbsurdCensor — Tue, 16 Jun 2026 15:39:02 +0000

At least for me, it's been pretty great, but I bought my system when it was $1800, now looks like the same system is $2700 and out of stock. I still haven't quite been able to run 120B parameter models under Windows, but for Qwen Coder 30B, it works pretty darn well for my at home needs.