Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Does that use a lot of energy?

hannahritchie.github.io

Post reply on HN
Print view View original

211–220 of 226 posts

Re: Does that use a lot of energy?

#211

Post by londons_explore »

cogman10 wrote: post #208
londons_explore wrote: post #181

Interestingly, the latest generation of LED lights are so efficient that it makes little financial sense to bother having a light switch. 330 lumens per watt from recent Phillips bulbs! The cost of having an electrician wire that switch is probably more than a little 2.5 watt light will use in it's lifespan - particularly when you account for the fact lights in hallways are probably in use most of the time anyway. Ad…

Do you have a link to the 300 lumens per watt lighting? I'm curious. I think that mostly only applies bulbs without inverters.

Philips UltraEfficient LED | Energy-saving lighting solution | Philips lighting https://share.google/lptcUJwHZWOQP1JMJ

They are direct AC bulbs, no inverter needed. They do have smoothing capacitors so they don't flicker, and I think a slight redesign of the drive circuitry could add ~10% more efficiency...

Re: Does that use a lot of energy?

#212

Post by dr_dshiv »

MichaelDickens wrote: post #35
dr_dshiv wrote: post #30

One hour of Claude code— well, I’d guess it would be comparable to an hour of driving an electric car. How to know?

OP says one query uses 0.3 Wh. Driving an electric car for 10 miles = 3,000 Wh which is roughly 10,000 Wh per hour. I'm not sure how many queries is equivalent to an hour of Claude code use, but maybe 5 seconds, which means an hour of continuous use = 216 Wh, or ~50x less than an electric car. OP has a longer article about LLM energy usage: https://hannahritchie.substack.com/p/ai-footprint-august-202...

My point is that Claude might easily be about 50x more energy intensive than normal ChatGPT prompting.

Re: Does that use a lot of energy?

#213

Post by ben_w »

noosphr wrote: post #103
NortySpock wrote: post #100

Earlier quoted context omitted.

Energy security is national security. Cheap electricity means you can do things that made "no sense" with expensive electricity. (e.g. smelt aluminum) Cheap electricity means you can underbid regions that have expensive electricity... As Technology Connections said, "Panels that cover your electrical needs for the next 25+ years? In the Midwest, we call that a good deal!"

Most of those technologies also need uninterrupted power supplies. Something wind, solar and batteries for the next 50 years aren't.

> Something wind, solar and batteries for the next 50 years aren't.

False. If you'd stopped before the "and" you would have been correct, though.

Batteries are really cheap now, and supply of batteries is growing basically as fast as people can get the investments and permissions for the inputs and the factories.

Re: Does that use a lot of energy?

#214

Post by ben_w »

supertrope wrote: post #203
Leno1225 wrote: post #200

Earlier quoted context omitted.

Yeah, or water towers. No need to play god here.

Pumped hydro energy storage relies on the cheapness of water and existing geology. If you have to build the chambers instead of damming a river it's too expensive. Most of the good spots to have a reservoir are already used. If you have to manufacture the bulk media instead of just using water it's too expensive.

Pumped hydro doesn't need a river, "just"* rock which isn't water porous and some nearby body of water (lake, sea, whatever).

The economics works out even if you were lifting concrete blocks rather than water, hence why you get pictures like this: https://en.wikipedia.org/wiki/File:Energy_Vault_Test_Tower_2...

The argument against lifting concrete is that you can dig a hole in the ground an pump water in/out of it for more reliability and lower cost than having a crane lift and lower concrete, and it's easy to make it much bigger both horizontally and vertically, so why bother.

But it does appear to be economical even with that, and water is cheaper.

We make lots of holes in the ground on a regular basis, including for extracting fossil fuels. Here's two, note scale bar, though I have no idea what the rock around it is like regarding water losses: https://www.google.com/maps/@50.9063171,6.4418046,17655m/dat...

* it's never "just" with things on this scale

Re: Does that use a lot of energy?

#215

Post by cogman10 »

londons_explore wrote: post #211
cogman10 wrote: post #208

Earlier quoted context omitted.

Do you have a link to the 300 lumens per watt lighting? I'm curious. I think that mostly only applies bulbs without inverters.

Philips UltraEfficient LED | Energy-saving lighting solution | Philips lighting https://share.google/lptcUJwHZWOQP1JMJ They are direct AC bulbs, no inverter needed. They do have smoothing capacitors so they don't flicker, and I think a slight redesign of the drive circuitry could add ~10% more efficiency...

Which of these bulbs? The highest efficiency I'm seeing poking around is 230L/W (Don't get me wrong, still very impressive)

Re: Does that use a lot of energy?

#216

Post by harrall »

b40d-48b2-979e wrote: post #194
dmd wrote: post #153

Earlier quoted context omitted.

Where does that 0.3 TW figure come from? That seems awfully high.

You're trying to converse with a LLM. It's made up.

Nope it’s 100% legit and I even remembered it wrong. It’s actually 0.6 TW.

You can get the number of TW from this report https://ourworldindata.org/energy-production-consumption

Convert TWh to TW for a year average.

Re: Does that use a lot of energy?

#218

Post by sha16 »

I wanted to write an app like this. Here's what I think would be useful for some people to compare: more checkboxes that represent typical daily activities like "how much energy does driving to the grocery store (x miles away) use" or "how much energy does idling in my car for 30 mins use", "how much energy does riding the bus/trolly/etc use", "how much energy does keeping my place warm vs kind of chilly", and also "how much energy does a cargo ship use to cross the pacific" or "how much energy does it take to source, manufacture, package, transport, and stock a pack of bubblegum". I think that would help people realize the largest components of their footprint and also how it relates globally.

Re: Does that use a lot of energy?

#219

Post by dmd »

harrall wrote: post #217
dmd wrote: post #153

Earlier quoted context omitted.

Where does that 0.3 TW figure come from? That seems awfully high.

Oops it’s actually 0.6 TW. https://ourworldindata.org/energy-production-consumption Convert TWh to TW for a year average.

Huh! I'm kind of stunned that we only use ~30x the power that we did back then. If I'd been asked to guess I would have added another 0 or even two of them.

Re: Does that use a lot of energy?

#220

Post by stratos123 »

dr_dshiv wrote: post #212
MichaelDickens wrote: post #35

Earlier quoted context omitted.

OP says one query uses 0.3 Wh. Driving an electric car for 10 miles = 3,000 Wh which is roughly 10,000 Wh per hour. I'm not sure how many queries is equivalent to an hour of Claude code use, but maybe 5 seconds, which means an hour of continuous use = 216 Wh, or ~50x less than an electric car. OP has a longer article about LLM energy usage: https://hannahritchie.substack.com/p/ai-footprint-august-202...

My point is that Claude might easily be about 50x more energy intensive than normal ChatGPT prompting.

A coding agent runs near-constantly, so of course it'd require a lot more compute than running even, say, a multi-minute query with a thinking model every hour. How much exactly is pretty hard to calculate because it requires some guesswork, but...

For a long input of n tokens from a model with N active parameters, the cost should scale as O(N n^2) (this is due to computing attention - for non-massive n, the O(N n) term is bigger, which is why API costs per token are fixed until a certain point and then start to rise). From the estimates from [1], it's around 40Wh for n=100k, N=100B. I multiply by 2.5 to account for Opus probably being ~2.5x larger than gpt-4o, and also multiply by 2 to pessimistically assume we're always close to Opus's soft context limit of 200k (it's possible to get a bigger context for extra cost, but I suspect people compact aggresively to not have to use it). That gets me 7.2J/t, which at a rough throughput estimate of 20t/s gives me power of 144W. Like a powerful CPU or a mediocre GPU, and still orders of magnitude lower than a car.

[1] https://epoch.ai/gradient-updates/how-much-energy-does-chatg...

Post reply on HN