Hacker News Reader

Comment by 37ef_ced3 | Hacker News Reader

Demand for GPUs for Ethereum mining goes to zero in Q1/Q2 2022, when the beacon chain merges with mainnet. Quote: "This will mark the end of proof-of-work for Ethereum, and the full transition to proof-of-stake."

https://ethereum.org/en/eth2/merge/

The price of (demand for) GPUs is strongly correlated with the market price of Ethereum. The Economist, June 2021:

https://www.economist.com/graphic-detail/2021/06/19/crypto-m...

Today, just like in 2018, we can see the price of Ethereum reflected in Nvidia's stock price. When Ethereum moves to proof-of-stake, mining GPUs will flood the market and demand for new GPUs will no longer be tied to the cryptocurrency bubble.

View on HN

More recently Vitalik has said Ethereum PoS is at least a couple of years away, likely due to new attack vectors that require further research https://arxiv.org/abs/2110.10086, one of which enables an attack with only 0.09% of the total staked. It would have been foolish to fork knowing attacks like these are still being discovered.

"with more than 99.6% probability, an adversary with 0.09% of total stake is in a position to execute a 1-reorg for any given day."

Another which could prevent the chain moving forwards indefinitely.

"an adversary controlling 15% of stake can stall PoS Ethereum"

I heard June-ish 2022, which is a couple quarters away, not a couple years. Could potentially be delayed further, but 3 quarters doesn't seem unreasonable.

https://eips.ethereum.org/EIPS/eip-4345

That's just a temporary delay, not when PoS would ship.

The recently discovered attacks are pretty bad and the mitigations are band aids.

If people believed that this is the end of PoW the prices of GPUs would already be going down, because people would know that they have at most 7 months of run (until end of Q2) and that they'd be massive selling of graphic cards. I don't see it happening which means that either:

* miners don't believe the switch is happening that soon

* miners know they'll keep PoW with some other blockchain

* miners are gambling hoping that they'll still make profit

Maybe all 3?

Maybe you don't realize the amount of profit that can still be made in those 7 months?

Miners earn 2 ETH every 15 seconds.

I'll believe it when I see it.

This. Ethereum already forked once. Ethereum GPU could remain its own currency for all of those miners who invested a shit ton of capital into GPU acquisition.

If the market value of Ethereum GPU is similar to the market value of ETC to ETH, the impact of Ethereum on the GPU market should shrink by 98%.

It could, but would it be worth anything? Miners only mine because they can sell the crypto to others for more than it cost to mine it. At current levels that would require quite a lot of buyers for this new coin with no significant advantages over the main branch.

It'd be worth less, but not entirely worthless. I think ETC is a good example of what happens on a deprecated branch.

Has Ethereum ever met a roadmap deadline?

Proof-of-stake was scheduled for December 2021, but it slipped to first-half 2022.

Don't kid yourself, it won't slip forever.

This is going to happen, soon.

It's going to happen at some point, I agree, but I wouldn't count on it by mid-2022, which is important if you are looking to buy a GPU. The fact that my 1060 is valued at 150% it's launch price indicates that miners are also pessimistic about the speed of a full POS transition.

Plus, I'm not convinced that the transition will go entirely smoothly. Obviously it would be in miners interest to prevent the POS transition, and if there's a bug or problems with the new chain I could totally see a fork or political pressure to delay things further (I don't think either is likely, simply possible).

Have you ever sought out to see which deadlines the Ethereum Foundation has setup and met? The only ones you'd read about online would be the ones that missed the deadline. But if you look at the previous milestones (like the Berlin upgrade), they all had their deadlines met correctly.

Are you telling us that one of the most popular cryptocurrency projects on the face of the planet which has access to thousands of the so-called 'brightest engineering minds' on the planet, after over 5 years, with billions of $ of funding at its disposal, has managed to make a minuscule change to its fee system? Wow, they're truly outdone themselves. What an exceptional team.

I wonder how many Ethereum developers it takes to change a lightbulb...

Many changes have been made, including a major upgrade to the fee system, with changes needing buy-in from many stakeholders. This is a network responsible for billions of dollars in value

Or maybe it's monopoly money, but critics always like to have it both ways, don't they. It's real when poor victims are losing it, but fake otherwise. It's all pretend, except when something goes wrong, then the teams running it are suddenly irresponsible.

Nothing ever goes wrong. A multi-million dollar hack is called a 'minor hiccup.' Missing a deadline by 5% of an entire lifetime is called 'a small delay' or 'a misunderstanding'. If they deliver within 10 generations, they will call it 'a resounding success, exceeding our wildest estimates.'

I think you vastly underestimate the difficulty on running a global network like that, with multiple parties who almost all have different goals and motivations.

I thought HN was supposed to be better than that. Changing the tweet length on Twitter is hardly just changing the column limit, there is so much more work behind the scenes. At least with Ethereum you could (if you're interested, doesn't seem like you are) see all the changes they had to go through in order to get that change on the network.

Total BS. I'm a blockchain developer and I worked on a project worth half a billion $ and we made far bigger changes than that in less than 2 years... With a far smaller team... Without even a single 'rock star developer'.

Either Ethereum devs are the dimmest minds in the industry or there is something seriously wrong at the foundational level of the Ethereum code itself.

> I'm a blockchain developer

Welcome to the club :)

> I worked on a project worth half a billion $

Wow!

> With a far smaller team...

There you go! Of course it's easier to change stuff when the scale is smaller. The difficulty of implementing something scales with the number of people working on it. If you're just one person (or a small team), changes are easy to work through. But if you're 100+ developers, multiple foundations and also private companies, even changes that are small will take a long time to go through the process.

> Either Ethereum devs are the dimmest minds in the industry or there is something seriously wrong at the foundational level of the Ethereum code itself.

Taking long time to change the core protocol is by design (it's a feature, not a bug), so everything is working correctly.

> Taking long time to change the core protocol is by design (it's a feature, not a bug), so everything is working correctly.

In that case, I hope it takes over 100 years to implement sharding, then at least the suckers who invested in Ethereum will rest in peace knowing that at least it met their decentralization needs after some other project has taken over the world.

Again you seem to not fully understand how things work, and you also seem to wanting to read my comments in the worst possible light. I never said it's designed to take as long time as possible to get out changes, I said it's designed to take long time.

But hey, you've already made up your mind so there seems to be little to no value of us even talking with each other.

Yep, Cardano can't get things running after years being flush with cash.

Well yes, I have paid close attention to some of the major deadlines. That's why I was asking here, to see if there was a major deadline they met that I missed. Because everytime I look at a roadmap, dates have slipped from the last roadmap.

It's pretty clear that Ethereum's team has a chronic problem with overly optimistic forecasting.

Has Ethereum put every deadline on indefinite hold?

It's actually got a testnet running now, so it won't be long before the merge and switch to PoS happens.

I'll doubt that'll ever happen. They've pushed up ETH 2.0 so many times that it's no longer believable.

Guess I'll hold off on the new PC build a little longer.

These prices are becoming the norm, not the exception, already took too long.. I am slowly giving up hope a GPU will go for MRP ever again. I also just want my GPU man

My decision is to not care about games or heavy number crunching/ML training. This frees me from needing a dGPU.

The added benefit is that my laptops get cheaper.

Right. Because the miners will stop their operations when Ethereum moves to proof of stake in 2030. Because there won't be a hard fork from miners who would rather have PoW. Because the miners will just give up on their operations when Ethereum isn't the shitcoin of the day. Because they won't move on to another shitcoin that is worth mining on GPUs.

> Because the miners will stop their operations when Ethereum moves to proof of stake in 2030

Well, from the perspective of Ethereum, they will, since Ethereum will no longer be PoW...

> Because there won't be a hard fork from miners who would rather have PoW.

There might be, or they might use Ethereum Classic. That's all fine, the second biggest network (Ethereum) will still be PoS so they can do whatever they want.

> Well, from the perspective of Ethereum, they will, since Ethereum will no longer be PoW...

Children develop object permanence by the age of two. How long does it take for cryptobros to develop GPU permanence ? You closing your eyes doesn't make the problem go away.

But the GPUs already exists, what solution do you imagine that involves the said GPUs to not be used anymore? Even if you sell them to normal consumers, the GPUs will be used, one way or another.

Are you suggesting we should destroy the GPUs that were previously used for Eth mining?

I think the point is that the miners already have the GPUs, they will probably switch their operations to mining other coins/forks instead of just giving up on everything.

Ethereum Classic fell victim to multiple 51% attacks in 2020. The hacker was able to gain control of 51% of the hash power and reorg some blocks and it allowed double spending and all the catastrophes that come with a 51% attack [1][2].

It is safe to say that it won't come anywhere near close to where normal Ethereum is in terms of mainstream adoption.

[1] https://www.coindesk.com/markets/2020/08/29/ethereum-classic...

[2] https://coingeek.com/over-1m-double-spent-in-latest-ethereum...

Miners will find another PoW ecosystem to deploy their mining hardware in, correct.

Ethereum is not really a shitcoin though, unless you consider all cryptocurrencies shitcoins. Many tokens are implemented as ERC-20s (including most relevant stablecoins), other chains target EVM compatibility to drive developer adoption, majority of the NFT boom has been in ERC-721 NFTs on Ethereum, and traditional companies are now building L2s on top of Ethereum tech (PwC, Reddit).

Ethereum blockchain is pretty fundamental to much of what is happening in crypto aside from pure NumberGoUpism. Miners can hard fork ETH if they want to, but I doubt they will. All those dollars held in stablecoins will determine which chain has validity, and I doubt USDC/DAI/RAI/GUSD would go along with a miner cashgrab fork again Vitalik...

>Ethereum is not really a shitcoin though, unless you consider all cryptocurrencies shitcoins

dingdingdingding! There is nothing useful done with Eth today, aside from quick money grabs.

There's a lot of people using it for online gaming (gambling) I think

Fueling addictions and bypassing regulations in hundreds of countries isn't exactly a shining example of the technology.

I'm no fan of it either, but plenty of folks would view bypassing financial regulations as a shining example. That was part of the point, was to sidestep government oversight.

Another way of saying that is "to enable criminal activity."

Sure. But legality is separate from morality.

It's important to separate moral judgement from technical judgement. Even if I agreed with you on morality (which I don't on either count), addiction industries, including the legal ones like tobacco and games, have always been large and profitable, and avoiding various regulation has also been a big "meta-industry" for a long time.

How come eth doesn’t use ASICS?

Also when they switch to pos how do they transfer your coins over?

Ethereum mining performance relies on memory bandwidth. Designing a GDDR6 or GDDR6X memory controller is hard.

I wonder how the M1 Max fares with its 400 GBps bus.

It won't do well.

Compare that 50 GB/s of the M1 Max to 760 GB/s of an RTX 3080. Then check prices. Even after ridiculous scalping markups, the RTX 3080 will still be cheaper.

> How come eth doesn’t use ASICS?

Ethereum was conceived to be ASIC-resistant from day one. They even considered making the coin resistant to GPU-mining but then changed their mind.

It's not that hard to be ASIC-resistant and seen that ETH GPU miners are still profitable to this date, they obviously succeeded in preventing ASICs from being economically doable to mine ETH.

> Also when they switch to pos how do they transfer your coins over?

Ethereum is already on PoS, but partially. At the moment mining is going on both chains: PoW and PoS. In a few months it's going to be PoS only. All the coins from the "old" PoW-only chain will still exist and be usable on the PoS-only chain.

> They even considered making the coin resistant to GPU-mining

This could have driven some very interesting developments in the CPU space, mostly in the memory interfaces.

"How come eth doesn’t use ASICS?"

It's intentionally designed to be memory-hard so that ASIC implementation is more difficult. Lots of people in the cryptocurrency world consider ASIC mining to be a misfeature, because it centralizes mining among firms that have the capital to do chip development. Litecoin's entire reason for existence is "Bitcoin, but can't be mined by ASICs".

"Also when they switch to pos how do they transfer your coins over?"

There's already a beacon chain running Ethereum proof-of-stake. Think of this like a dark launch for web tech - basically you run the new infrastructure in parallel with the old infrastructure, make sure it's stable, get everybody adopting it, and then cut over the UI to use the new infrastructure.

There's a one-way bridge that ports state over from the PoW chain to the beacon chain. Think of this as a double-write layer; all new transactions get written both to the PoW and PoS chains.

In early 2022 "The Merge" happens. This takes the form of the "difficulty bomb" - basically, Ethereum mining will get exponentially harder and eventually unprofitable until all the existing miners turn off their machines and switch to PoS, at which point there will be no point to mining Ethereum and no point to writing transactions to the PoW chain.

Thanks. I had assumed PoS was more like a holy grail for cryptos? But this sounds like it’s entirely possible?

Will PoS lead to a run on other resources like hard drives?

No. PoS is secured by miners (called validators in PoS) putting up collateral. It takes 32 Ether to run a validator. If they act maliciously that collateral gets burnt.

Nah, hard drive usage is about the same with PoS.

It could potentially lead to a run on Ethereum. The resource used in PoS is the cryptocurrency being secured itself. Basically, you agree to lock up a certain stake when you run a validator, and forfeit it if it's found that your validator "cheats". By doing this, you make cheating unprofitable - there's a financial incentive to catch cheats, and a financial disincentive to cheat, and the cheaters eventually go bankrupt (or end up on their own isolated chain where nobody transacts with them).

The financial effect of this is that a bunch of Ethereum gets locked up and illiquid. Instead of miners receiving Ethereum which they want to sell for dollars to run their hardware, stakers lock up Ethereum and effectively receive interest. This reduces immediate supply within the market. I'm wondering if a lot of the recent price movement is in anticipation of this.

Why didn’t they create more eth as part of this change to offset that issue?

Also does eth supply increase forever or does it have a hard coin limit like btc?

ETH stakeholders want ETH to go up so a change that minted a bunch more ETH in relation to PoS would probably be contentious.

Currently no hard cap on issuance, 2 ETH are minted every block, but ETH burn rate is expected to outpace issuance post-POS.

Eth keeps getting generated forever but at the same time a portion of transaction fees gets burned, so whether coins in circulation grows depends on transaction volume and fee levels.

It's complicated, and basically comes down to Vitalik's observation that "Ethereum is a currency in service of a protocol, while Bitcoin is a protocol in service of a currency." Basically the Ethereum devs set the economics to incentivize the correct behavior from players in the ecosystem, and this can be either inflationary or deflationary depending on gas prices, network usage, amount staked, etc. It's responding to the supply & demand for computation on the network.

For most of its history Ethereum was inflationary. It flipped and became deflationary a couple months back, and not coincidentally the price started rising again. It's likely to be significantly deflationary under Eth 2.0 with expected transaction and staking volumes.

And the price of ETH is strongly correlated with the price of BTC. So we will see if those GPUs flood the market or simply switch to another crypto.

Reply on news.ycombinator.com