Topic: Why will GPU mining be useless? Difficulty/Price relations? (Read 4492 times)

The big thing with Avalon chips is your in the ASIC game for $150ish per 4.5GH/s you don't have to put down a ton of money.

Price, as with any market, is determined by supply and demand at the combined market venues. I cannot support the argument that demand is increased based on increases in hashrate. It's not like people would look at the hashrate chart and say "Wow, hashrate has gone up! I need to buy some coins!" It's much more likely that buyers and sellers watch price and react to news, price movements, etc.--exactly the same forces that move stock prices every day. There is herd behavior, buying and selling panics, etc., just like any other speculative market.

Miners are not really a market force--they react to the profitability and turn on miners when profitable and turn them off when not. There is some feedback here; turning miners off and on influences hash rate and difficulty directly. As far as the coins mined...some miners hold, some sell...some mine AND even buy more (like me). So they are part of the market...or not, depending on their goals.

I just can't buy the argument that difficulty increases cause price to rise. Look at May 1 to now. Price has gone up and down post-bubble; difficulty went from <10,000,000 to over 18,000,000 (in a few days), an 80% gain. Price has essentially stayed unchanged--not reacting at all to 1.5% average daily increases in difficulty for 6 weeks.

ASICs will be profitable at nearly any price for a long time; there will be NO stagnation or decline in difficulty for the next year or so--it will simply keep rising, barring BTC price dropping under $10 or so. I can't see how that is going to influence price. On the contrary--all of the ASIC vendors accepting bitcoin for payment will be cashing them out (yes, buyers will be buying BTC to pay for hardware--but I'm betting that many ASIC buyers are putting existing BTC to use), which is downward pressure on price.

Supply and demand...that determines price, like any other market.

I believe they are correlated, perhaps even to the point of codependent.... a bit of the chicken/egg conundrum... price is affected by difficulty, and difficulty is affected by price. Discerning difficulty from hash rate is just ignorant, difficulty is defined by hash rate, I just took the extra step out of the equation in the OP.

As for them being a coincidence, no I do not think so.. Coincidences are things that happen to occur simultaneously whilst having no codependency or correlation.


EDIT:

I realize that systematically, that price and hash rate have no relation... a price is not explicitly set by how much power is on the network.. that's obviously implied

I think you stopped reading when I mentioned that unrelated correlation is the definition of the word "coincidence".

 I also think it is possible that hash rate and price influence each other (in other words, the causal relationship goes both ways), but I haven't seen a good case made for hash rate influencing price.

difficulty follows price and not the other way around. people will start mining because of the price increase, afterwards difficulty will adjust.

same way if the price decrease--> people will stop mining--> difficulty goes down

Can you list any other coincidences that have been going on for 4+ years with predictable results?


I'm not trying to say that network hashrate specifically dictates BTC trade price (last month's bubble is an obvious example) but to say they aren't related is just bogus.

Erg ya got me!

True, but I can't find anyone to buy from   damn regulators..  was just about to gox some money to bits

Simply buying the bitcoins instead of mining them would give you a better return in this case.

Unless BTC keeps rising.  An investment in mining this year that doesn't cover electricity cost could make a very good return in a couple years.  At some point we could be looking back wishing we were doing processor mining up to the point ASIC's came out, much less GPU's.

Now, for the future of GPU mining...

The point was made that even though GPUs will be able to mine fewer and fewer bitcoins, the rise in the price of bitcoins will continue to enable profitable GPU mining. Unfortunately, that is probably not true because of the economics of the mining market. The mining market is (will be) nearly perfectly competitive (http://en.wikipedia.org/wiki/Perfect_competition). That means that operating margins will be very small. As a result, only the most cost-efficient mining will be able to compete, and GPU mining is much less cost efficient than ASIC mining. I could do the math, but I think it is pretty clear, and I am lazy.

On the other hand, GPU mining will remain feasible where electricity is cheap or free. Also, people might continue to mine with GPUs for fun or for some other reason even if they are losing money.

You are confused. The word you are looking for is "causation". You want to say that correlation does not imply causation. In other words, the fact that A and B are correlated does not imply that A causes B or that B causes A.

Price and difficulty are highly correlated. Just look at the graphs. It is pretty obvious. Keep in mind that totally unrelated things can be correlated (that's called a coincidence).

Anyway, it is easy to make the case that price influences difficulty: If the price goes up, more people mine and the difficulty goes up. If the price goes down, less people mine, and difficulty goes down.

It is not so easy to make the case that difficulty influences price, though you could make this case: difficulty and price are both influenced by interest in bitcoin.

I think this derailed the conversation.  It looked like he was arguing that GPU's would still be quite valid in the future as the price of a BTC rises.  If we were still at last years prices with today's difficulty GPU's wouldn't be worth it.  We never would have gotten this many people putting their processing power in without the rise in BTC, and couldn't have gotten to this difficulty.  The price has a lot to do with total network hashrate.  As ASIC's come in, the future of the GPU will be determined by the price of a bitcoin, like he said in the first post.  There are a lot of people with a lot of bitcoin who want it to succeed, the price will rise.  GPU's future look fairly rosy.  

New gpus can be paid for in 6 months at CURRENT difficulty, which will not last. Contrary to your contrary there is no correlation between price and difficulty, there is a correlation between difficulty and price ( in this case order matters ). This means that if difficulty goes up price doesn't change.  GPU will be run out by ASIC. The only question is how long til then.

Yes, it requires some knowledge, but Avalon is basically providing the circuit specs, it's just the layout that needs to be done. There are plenty of companies that will run your board design to whatever quantity of boards you want (the more, the cheaper). That's definitely a non-issue. I'm certain there will be at least two competing designs ready in time for the chips. Populating the boards might a bit more problematic--SMT parts are not exactly DIY-friendly.

A lot of people purchased Avalon chips hoping that they'd be able to do something with them when they arrive.

I suspect a lot of these folks are going to be disappointed to learn that there are very few legitimate options to turn their chips into a mining machine, and the ones that do exist will either be very expensive or have a huge backlog.

Designing and manufacturing circuit boards is not a walk in the park, and requires some serious technical know-how combined with very expensive machinery.

We'll see...

GPU's will be good for a while still. New GPUs can still be paid for in 4-8 months.

After that, it will depend on your utility prices and the BTC exchange rate.

Contrary to what was said earlier, there is (currently) a very good correlation between difficulty and BTC exchange rate. That means that, even though GPU's will mine a lot less BTC, the BTC will be worth more in fiat terms and this will probably offset the dip in BTC mined (so you can still pay your bills and keep your GPUs running).

This is of course an educated guess, as everything is in flux...

It's not as hard as you think to make an ASIC miner with Avalon chips. And a lot of people are working on it.

Once Avalon batch 3 ships and BFL begins to ship in quantities, and ASICMiner deploys the 200TH/s worth of mining equipment they have, you'll see ASIC prices come way down because it won't be profitable anymore. GPU mining will be long dead by then. As dead as CPU mining has been for a while now.

Bullshit.

Avalon batch 3 isn't shipping yet, and they cost 75 BTC (~$9000). Given that batch 2 is just now shipping, it looks like batch 3 owners still have to wait a few weeks.

It is a big jump from buying discrete Avalon chips to getting a working ASIC miner. Has anyone actually done this yet besides Avalon? I think not.

BFL has yet to ship a 60 GHs machine to anyone. As far as I know, all they have done is shipped a handful of 5 GHs machines.

They do indeed have incentive to sell -- particularly, to pre-sell. Designing and fabricating ASIC hardware costs months and millions, and Bitcoin is a pretty tiny market. I think it's an "avg" block mining duration of 20m, right? 25BTC*3=75BTC/hr*24=1800BTC/day. Multiply by current BTC price ($120-ish), and you get $216k/day. It's a pretty small market, and not too many people are going to want to make multi-million thrive-or-die bets on producing a boat-load of ASICs -- certainly not hardware manufacturers. Allowing pre-orders (whether the hardware or allowing would-be consumers to purchase equity in the company) is one of a small handfuls of way these designers/manufacturers have any shot in Hell of being able to produce these devices.

Consumers basically get a discount for extending a loan to the company, and we've seen how that is good, and most prominently in BFL's case, how that can be very bad. In the near future, if manufacturers can just snap their fingers and have 100TH/s of ASICs within a week without needing third-party financing, then they'd have no incentive to sell for a rate that's profitable both for manufacturer and consumer.

You can't buy it because you missed out. When Avalon ships batch 3, there will be 100 TH/s of Avalon machines out there.

Or you can buy discrete Avalon chips for about $9.50 for 282 Mh/s. That is $2200 per 65 GH/s. Even taking into account cost of the PCB and power supply, that is still in the same range.

Or you could have ordered BFL for about $1300 / 60GHs.

ASICMiner is more expensive, but it's only for the people who have missed out. Their prices will come down when the Avalon chips ship and you see DIY Avalon machines anywhere.