Nvidia Limits RTX 3060 Hash Rate • Hacker News Board

Re: Nvidia Limits RTX 3060 Hash Rate

#501

Post by pyrale » Mon, Feb 22, 2021, 1:15 PM UTC

mrb wrote: post #488

arcticbull wrote: post #468

Earlier quoted context omitted.

> Again: miners expend the same amount of energy regardless if they are validating 1 or 1000 transactions in a block. For one that’s horrifyingly wasteful - you say it like it’s a good thing but it’s really not. And two, I’m describing the network as it exists today. That’s not dishonest; there’s no plan to increase the quantity of transactions per block. If it ever changes we can run the numbers again. But saying it…

« For one that’s horrifyingly wasteful - you say it like it’s a good thing but it’s really not. » It is a great thing: it means Bitcoin can scale without increasing energy consumption. See, that's my problem with the way you phrased your post. You misrepresent the system. You imply transactions consume energy when they don't.

Not really. It means Bitcoin _currently_ scales negatively on market cap rather than transactions. Considering that bitcoin seems to be advertised primarily as a store of value and that there is little effort to push adoption for transactions, that is a very bad thing : it means that, in case of success, bitcoin performance per transaction will become worse.

In the long term, if transactions count doesn't rise fast, we will also see significant issues, either with transaction fees going to absurd amounts, effectively banning btc as a payment medium, or with the miner pool decreasing to counts where decentralization becomes inexistent.

Re: Nvidia Limits RTX 3060 Hash Rate

#502

Post by zaarn » Mon, Feb 22, 2021, 1:43 PM UTC

mseidl wrote: post #446

sellyme wrote: post #305

Earlier quoted context omitted.

> The oddest solution I can think of is gamers lease graphics cards at a low low price and nvidia mines on them in the off-hours. This would be a complete non-starter in markets with high electricity costs or temperatures. Running a GTX 3090 at 100% load during the summer where I live would cost $140/month and will heat up my room to about 65°C.

This is the other thing I don't understand. I'm from Germany and our power is around 35 cents per khw. Even using a 14th/s asic with bitcoin as high is it is now, you're losing lots of money. It uses 1,3 kilowatts. The fact that the prices of the cards are so high in Germany now too... the 3080s are at 2000€ plus.

Bitcoin is certainly not an option in Germany. Other mining operations may be borderline profitable, if you have a card already lying around.

Re: Nvidia Limits RTX 3060 Hash Rate

#503

Post by dahart » Mon, Feb 22, 2021, 3:10 PM UTC

MavropaliasG wrote: post #470

dahart wrote: post #385

Earlier quoted context omitted.

Sorry, you're right, I misunderstood the Linus reference as Torvals not Linus Tech Tips. But so you count reductionist click bait opinion commentary on YouTube as "evidence"? Commentary that comes from a company who's profit motive is driven by clicks and attention? Saying that companies primarily care about their bottom line is tautological, it contains no information. It's still a straw man regarding whether or not…

Did you watch the damn video mate?

Yes, I did, twice. The arguments in the video do not contain a single point that hasn’t been covered by people repeating the talking points here.

The thing Linus didn’t address, and the thing people don’t seem to understand about the basic economics of the situation is that there is no amazing second hand market getting flooded with cheap GPUs without first getting hosed by miners, having the prices go through the roof and not being able to get one for the next year. If that happens, then yeah, sure, the used ones will be cheap some time later, and you don’t get to choose when or even if mining crashes before the next GPU update. If that’s really what you want... enjoy.

That video is primarily an ad. It’s an ad for a VPN service and literally a ball-shaving kit. “If you imagine it any other way, then congratulations, you played yourself.” - Linus.

Re: Nvidia Limits RTX 3060 Hash Rate

#504

Post by mywittyname » Mon, Feb 22, 2021, 3:47 PM UTC

wruza wrote: post #498

ddtaylor wrote: post #480

Earlier quoted context omitted.

I think the market is working as expected here, since a higher price is placed on the ability to decentralize finance versus having a few people playing video games at higher quality. If you're interested in machine learning there are resources specifically for that, such as Google's Tensorflow Processing Units (TPUs)

“higher price is placed on the ability for few people to participate in speculation and money laundering versus having better 3d modelling in realtime for everyone” This word play can be played backwards too.

People who use tools to make money are going to spend more on them than those use use tools for entertainment. So if a company wants to prevent pros from buying their consumer-grade products, then preventing the consumer-grade tools from being used for pro features is a great way to do it.

nVidia has been doing this for decades with their Quadro cards. They use the bios/drivers to hide the features necessary for CAD out of the GeForce cards.

Re: Nvidia Limits RTX 3060 Hash Rate

#505

Post by derefr » Mon, Feb 22, 2021, 3:52 PM UTC

jkilpatr wrote: post #324

derefr wrote: post #203

Earlier quoted context omitted.

IMHO the most important endpoint for ETH2 isn’t ETH1 under the beacon-chain, or greenfield projects under the beacon chain, but rather major projects that have already been in development for years, and already had a testnet targeting some alternative sub-chain substrate (e.g. Polkadot, Near), re-evaluating their alternatives at mainnet launch time, and choosing to use ETH2 as their mainnet substrate chain instead. T…

> and choosing to use ETH2 as their mainnet substrate chain instead. But why exactly? Until ETH1 is in the ETH2 chain ecosystem ETH2 is just another blockchain platform, providing no network effects. ETH2's big selling point was higher throughput but bridges are already starting to commoditize access to more throughput from ETH1. ETH2 could provide smoother access than the bridge interfaces provided by other chains.…

The difference is that ETH1 will bridge to ETH2 in the future, and therefore by choosing ETH2 as your substrate, you’re predicting that transaction costs for sibling-transactions crossing from your chain to ETH1 will become cheaper when that happens.

A similar promise cannot be made for any other substrate network, as ETH1 isn’t going to bridge to any other substrate network.

Since it’s not like sub-chain mainnets can pack up and move house after they’re launched, when launching one, you have to make choices based on how things will be, rather than how things are. It’s sort of like choosing a city to build a corporate headquarters in, based on what the civic infrastructure is likely to look like in 20 years.

Re: Nvidia Limits RTX 3060 Hash Rate

#506

Post by nottorp » Mon, Feb 22, 2021, 4:26 PM UTC

michaelt wrote: post #152

Netcob wrote: post #113

Earlier quoted context omitted.

It's done so that when the crypto bubble bursts again, the market isn't flooded with used graphics cards. In 2017 everyone was buying GTX 10XX cards, on one hand because they actually represented a good improvement over the last generation for a change, but mostly due to the crypto bubble. Which then burst at the beginning of 2018. Then RTX 20XX came along, which actually represented a slight drop in price/performanc…

And at the same time, it lets them cut out the scalpers and sell straight to the miners who'll pay twice the price for a card - without the price-gouging upsetting gamers.

But I'm a gamer and upset already :)

I just checked and my 1050Ti costs new about 30% more than what I paid for it 2 or 3 years ago...

Re: Nvidia Limits RTX 3060 Hash Rate

#507

Post by nybble41 » Mon, Feb 22, 2021, 4:58 PM UTC

sandworm101 wrote: post #168

x86_64Ubuntu wrote: post #165

Earlier quoted context omitted.

Why is Free Speech necessary to a Free Market? I thought a consumer choosing to spend their dollars was the "speech" in a Free Market environment. And when you lobby the government, you've kind of thrown out the Free Market paradigm.

Free speech is also product review, such as LTTs recent video on this subject. Regulation of a market by government doesn't mean abandoning free markets. Regulation is an essential part of sustaining a free market. Totally non-regulated markets quickly devolve into monopoly and other anti-consumer evils such as devices, like this, designed to kill secondary markets.

> Regulation of a market by government doesn't mean abandoning free markets.

"Regulation" in the old sense of "making regular", perhaps. Establishing fair weights and measures, preventing fraud, etc. Ensuring that individuals are free to trade and that all transactions are in fact voluntary (including, most importantly, transactions with the government itself). But when your "regulation" consists of applying force to micromanage how people—shareholders in this case—employ their capital for the benefit of some other group, you've moved well outside the realm of the free market.

Re: Nvidia Limits RTX 3060 Hash Rate

#508

Post by nybble41 » Mon, Feb 22, 2021, 5:21 PM UTC

throwaway-8c93 wrote: post #281

ryebit wrote: post #196

Earlier quoted context omitted.

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…

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?

> 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.

Re: Nvidia Limits RTX 3060 Hash Rate

#509

Post by rtkwe » Tue, Feb 23, 2021, 1:56 AM UTC

mrb wrote: post #421

autocorr wrote: post #390

Earlier quoted context omitted.

Can someone explain how transactions are not insanely cost prohibitive for Bitcoin with power consumption figures like those? Those numbers would suggest something like >20 USD. Is it because most transactions in practice actually happen internally on exchanges that avoid putting every individual transaction on the blockchain? I have limited knowledge of Bitcoin so maybe not quite "explain like I'm five"... but close…

« Can someone explain how transactions are not insanely cost prohibitive » Because the comments above yours are misleading. Transactions don't consume mining energy. Miners expend the same amount of energy regardless if they are validating 1 or 1000 transactions in a block.

So essentially it only gets worse from these numbers, because right now every block is full so every transaction is consuming the least amount of energy it can if you process less transactions the energy cost just keeps going up.

Also the defense of "the energy is spent securing the blockchain" is missing the point, the blockchain isn't a good unto itself it's only useful as a log of transactions and you only need new blocks when there's transactions and if there are no transactions Bitcoin is basically dead.

Re: Nvidia Limits RTX 3060 Hash Rate

#510

Post by tuankiet65 » Tue, Feb 23, 2021, 7:34 AM UTC

autocorr wrote: post #448

tuankiet65 wrote: post #394

Earlier quoted context omitted.

The whole point of Bitcoin mining is to produce a block , which contains a bunch of transactions. The Bitcoin network dictates that each block's SHA256 hash starts with a certain number of zeroes, so the only way to achieve this is to brute force the block data until you find the "winning" block which hashes to a number of zeroes. Once someone finds a winning block, they're rewarded with a number of Bitcoin. This sub…

Thank you! That was a really helpful response. That makes a lot more sense that many transactions are included within a block. So effective transaction fees are likely in the few cents USD range. I've read that one issue with the latency of Bitcoin transactions is that many of the large miners from China have network issues. Would raising the block size to 10-100MB cause significant latency issues for transactions ev…

> Would raising the block size to 10-100MB cause significant latency issues for transactions even if one could fit a lot more transactions into a single block?

There are two ways to understand your question so I'll just answer both of them:

Fitting more transactions in a block leads to faster confirmation, and potentially lower fee. When a transaction gets into a block, the network confirms that the transaction is valid and can't be reversed. Right now each block can only fit ~2k transactions, and since a block can only be generated every 10 minutes, the mempool (backlog of unconfirmed transaction) is rather huge. This site shows how big it is: https://jochen-hoenicke.de/queue/. Miners are incentivized to only select transaction with high fee into the next block, which in turn incentivizes users to pay more fee to get their transaction to confirm faster. With bigger blocks, more transactions get confirmed quicker, which leads to lower fee.

Bigger blocks however mean miners with slow internet connection are at a disadvantage. When a miner finds a winning block, they broadcast the block to as many nodes in the network as possible. After a while, a majority of nodes accepts that winning block, and the miner is eligible for the mining reward. If A finds the winning block 10 seconds after B, but A is able to propagate their block quicker than B, then it's possible that A's block is accepted and not B'. In reality, it's possible that B might still win the race, but the rule is: faster propagation => more chance of being accepted.