I think specifically it's comparing gas heating vs AC. Heat pump heating would probably do better. In other words, it takes less energy to move heat inside/outside than to "create" it

(With caveats like heat pumps are much less effective in extreme cold)

You can select gas heating vs electric heat pump. The heat pump looks to be about a third the cost of gas.

Definitely not UL (if we're talking about the $300 inverters on Amazon or AliExpress from companies like Anenji)

I thought Discord bans people for using alternate clients?

Deye/Solark famously bricked a bunch of inverters remotely not long back

I opened up all inverters here and disabled all remote access hardware physically. I'd rather not have a warranty than the risk of either seeing my critical hardware hacked or bricked by the manufacturer. And given the size of these (one is a 17KW unit) the risk of real damage to my home circuits and/or the feed is not imaginary, just rapidly turning it off and back on would already be a big problem, especially on a really sunny day. I try everything I can to not trip the breakers on the inverter around solar noon on sunny days because the amount of current is non-trivial, even on a 960V system. It will generate an absolutely spectacular arc.

My SolArk inverter came with a wifi dongle that uses Modbus over RS232 to interface with the inverter. I removed the dongle and connected a Pi to provide a Modbus-over-TCP interface that I can use from Home Assistant. I got more control (the officially-supported Modbus interface is read-only), and SolArk no longer has any access to my equipment.

That's a nice trick. I should try that.

A lot of the *claws emphasize binary size and lines of code. I think for better or worse people treat codebase size as a proxy for "how much of the project is unsupervised, unmaintainable, buggy AI slop?"

It seems to support connecting to your own LLM on the same LAN

The point is the agent is still the LLM. No LLM, no agent.

LLMs are not agents. LLMs are language models that simply respond to a text prompt with a textual response. Agents are middleware that take input from the user and then use LLMs to drive tools.

They are just a to-do list. The real work is done by the LLMs

An LLM has no motive power, like a script without an a cast, or a program without a computer to execute it.

I tried connecting OpenClaw to ollama with a V100 running qwen3.5:35b but it was really, really, really slow (despite ollama itself feeling fairly fast).

These "claw" agents really multiply the tokens used by an obscenely huge factor for the same request.

i recently decided to get into this ocean boiling game too, the 32GB V100 seems like a pretty good VRAM/$. if i may ask, do you make any special accommodations for cooling? i've never dealt with a passively cooled card before and i'm curious whether my workstation fans (HP Z840) will be sufficient. i'm going to try 2 cards at first but i think i might be able to squeeze a third in there

No. I have a Titan V CEO edition, which is basically a 32GB V100 but has full active fan cooling which I'm finding works just fine.

Oh very cool. Some folks are printing shrouds for dual 40mm fans so I'll probably try that if the stock case fans don't do it

Maybe a dozen or so? But if you only have 3 bags of groceries you can just use a regular bike + basket.

The city of LA runs its own asphalt plants, and there's a bunch of other companies selling it in LA County. A lot of the raw material is recycled from roads that are being torn up, AsphaltPro had an interesting article about it recently

Most EVs are SUVs

(x) to doubt