News for hackers. Stuff that matters.
Re:Just 137 crypto miners use 2.3% of total U.S. power by pcthrowaway on Friday February 09, 2024 @03:12PM (#39319925)
Different PoS chains will have different characteristics, but I was curious about this claim so I dug up some numbers for Ethereum, Bitcoin, and Visa:

Transactions in 2023:

    Bitcoin:    174310000  
    Ethereum:   583530000  
    Visa:    212600000000
Energy consumption:

    Bitcoin:  167.8          TWh/yr   
    Ethereum:   0.0067479963 TWh/yr  
    Visa:     197.57         TWh/yr (2021?)
Energy consumption/tx

    Bitcoin: 962.652745 kWh
    Ethereum:  0.011564 kWh  
    Visa:      0.929304 kWh
It's entirely possible some of these numbers are wrong or off so please check my math, but it looks like Ethereum is the most energy-efficient of the 3 on a per-transaction basis, maybe an order of magnitude better than Visa.

Bitcoin on the other hand is 3 orders of magnitude worse than visa on a per-transaction basis. I suspect both Bitcoin and Ethereum also transact more value in USD per transaction (if you're willing to consider funny crypto money at spot prices) on average, than Visa, by virtue of them having a much higher transaction fee which is also borne by the address making the transaction.

If you don't consider Ethereum transactions to be "effective work" than this isn't a relevant metric of course, but I'm curious how you're arriving at your figure.

Re:Just 137 crypto miners use 2.3% of total U.S. power by FabHK on Friday February 09, 2024 @11:54PM (#39323708)
> curious how you're arriving at your figure

Hi, sure. The nodes in the EVM do not collaborate to split the work, rather, they all perform (basically) the same work. So, you could instead run Ethereum with one node that connects to the net and receives transactions, processes them (around 20 per second - it's not exactly a lot of processing), puts them into blocks, and sends out the complete blocks. It would perform exactly the same as the current network, and would consume, say, 4 W (a Raspberry Pi, say).

The one difference is that it would have to be trusted, which is the big no-no. But that's the effective work being done.

But the validators burn around 1 MW (that lines up with your figure approximately) and are paid $5m a day to do that work. And 1 MW / 4 W is a factor of 250k, which I rounded up to 1m considering all the other activity around Ethereum, though feel free to round it down to 1e5.