[ryebit]

I think right now biggest earner for GPU mining is Ethereum. And (I think) it's aiming to move to Proof of Stake sometime in next 12-24 months, which kinda time limits the utility of both Eth-specific mining chips, and large investments in GPU mining (unless mining burns GPUs out really fast?).

Regardless, most proof of work algorithms are already being optimized with ASICs (ala bitcoin)... Or, like Ethereum's ETHASH alg, they have a step requiring high throughput random access to large amounts of memory, ala scrypt or argon (5gb or so for ETHASH right now, and slowly increasing). That's what's kept ASICs from being profitable for Eth, and is probably limiting factor for FPGAs as well. Though maybe there are FPAGs with dedicated memory on par with a modern GPU?

[kortex]

Proof of stake is the fusion power of crypto. It's been 6-24 months away since what, 2016? I toured Consensys a few years back and at the time it seemed PoS was right around the corner.

OTOH, it could be cracked tomorrow and value of dedicated silicon will go off a cliff. So I don't blame the GPGPU crowd for that, before even considering the engineering challenges of silicon/FPGA for eth.

[ryebit]

Ethereum's Proof of Stake "beacon chain" is already live, with around $6 billion staked (since launch in december 2020). Next year is just formalizing and testing it taking over from the miners. So despite timeframe, there's little uncertainty about the outcome.

https://beaconcha.in/ is a nice explorer to see beacon chain running live.

[kortex]

Neat, thanks!

[throwaway-8c93]

A question out of genuine curiosity: Doesn't the nonce puzzle difficulty adjust to the amount of compute the network has, making it a zero sum game? If every mining pool buys 20% extra GPUs, isn't everyone soon back where they started, wasting hundreds of millions in the process?

[ryebit]

I'm not an expert at the mechanism design, but yeah, that's my understanding. The network's goal is to have a block every X seconds; and the difficulty periodically adjusts if blocks come in too fast / too slow (due to hashpower coming on/off line).

(Ethereum & Bitcoin both add some other year-decade timeframe factors to the difficulty, but don't think it affects this basic principle, which acts on week-to-minute timeframes).

I view it as a race to the margins. Anyone who can do it cheaper, or benefit from economy of scale, will get a larger slice of the pie, but at smaller margins... creating a cycle of consolidation.

I think original assumption was that folks running PoW at home on their GPUs would be able to compete, but due to efficiencies of specialized hardware and regional differences in electricity costs, that's just not the case.

---

Others might disagree, but my personal opinion is that this isn't a sustainable way to maintain decentralization, since it seems to obviously trend towards a few large "just turning a profit" players.

There might be some ways to adjust mining incentives to make it work, but this fundmental issue is why I think Proof of Stake has a much more viable future, as it sidesteps this (and the environmental) issues.

[nybble41]

> If every mining pool buys 20% extra GPUs, isn't everyone soon back where they started, wasting hundreds of millions in the process?

In theory you could run the entire network on one CPU, though it would take a very long time for the difficulty to adjust downward by that much from its current level. However, the extra capacity isn't wasted—a network secured by the hashing power of a single CPU would be extremely vulnerable since a trivial expenditure would give an attacker >99% of the hashing power and the ability to create forks and double-spend almost at will. The point of spending so much on hashing is to ensure that no attacker could plausibly out-spend the rest of the network and thus acquire a majority of the hashing power.

Now, do we need all of the current hashing power to secure the network? I don't think so. That's a consequence of the initial block distribution system, with a fixed halving schedule for block rewards and a price which has risen much more quickly than anyone anticipated. The cost of reliably executing a double-spend via a 51% attack far exceeds the potential reward for pulling off the attack. At some point the market will be saturated, the price will cease to double every four years (or less), transaction fees will make up a majority of the block reward, and the resulting drop in the total market value of the block rewards will lead to a decrease in hash rate to a level which is sustainable and yet still sufficient to discourage any would-be attackers.

[galangalalgol]

So nvidia is releasing a line of crippled cards that will mostly only be used for one year amd then trashed? That is horrible! Glad it may be over soon I guess.

[ryebit]

Pretty much, from what I can tell!

I sorta see this as Nvidia trying to create some market segmentation not for the miners, but so gamers can actually get their cards before Nvidia loses mindshare. (Not that I think that's likely)

[csunbird]

GPUs have a resale value on second hand market, ASICs or FPGAs usually do not, unless sold to other miners.