Latest Shipment of Avalon ASICs Could Increase Network Hashrate By 500% | Hacker News Reader

Latest Shipment of Avalon ASICs Could Increase Network Hashrate By 500%

thegenesisblock.com

46 pointscgi_man45 comments

ASICs are going to rid a large amount of pools. GPU mining and FPGA mining will fade out, because the value returned will be so minimal compared to these monsters. There is a finite amount of bitcoins released for a set amount of time. The power consumed/bitcoin return ratio on GPUs/FPGAs as the difficulty rises due to the ASICs will just leave people frustrated.

GPU mining had one benefit. Once you gave up, you could at least resell the video card or play games. ASICs are basically trash. The FPGAs can be recycled on ebay.

For those uninitiated: Assume I am on an i7 2600k and you are on a Pentium 2 clocked at a fine 200mhz. If we are mining in the same pool (just us), I will probably do 99.99% of the processing, and receive that return. You probably spent more power than me with your outdated technology, and only received 0.000001% of the payout.

I think the tech has gone CPU > GPU > FPGA > ASIC.

Makes me wonder what will come next.

You do realize the whole premise of an application-specific-integrated-circuit is to do one thing the best. If properly architected, an ASIC that does the most SHA hashing with the least transistors will be the best miner, inherent to the technology.

That doesn't mean the designs cannot be improved... And once somebody releases an improved design, the older ASICs become far less useful.

At this point, an "improved" design would have to mean algorithmic improvements to a hardware SHA algorithm (i.e. not likely).

Improvements that you can actually count on would be die shrinks, but these will provide returns in accordance with the tail end of Moore's law. Compared to the multiple-order-of-magnitude improvements from the CPU -> GPU/FPGA -> ASIC jumps, these improvements likely won't even be worth writing home about.

Maybe someone will figure things out at the circuit level to get hashes out faster. Or figure out better ROMs (with fewer transistors) to store the 64 32-bit fractional parts of cube roots of primes. Or, someday, someone will come up with the coinotron, which, like a transistor or memristor, would be at the boundary between chemical and electrical engineering, targeted at the bottleneck in the hash.

There is no next step. ASIC is the end-all be-all.

(Not even quantum computing can help - there is no known algorithm to help find SHA256 preimages, which is what would be necessary for a Bitcoin mining breakthrough.)

ASIC is the 2nd to last step. The next (final) step is a full custom design, co-developed with the fab process. This is how Intel & AMD design CPUs.

Well, standard cell ASIC vs. full custom ASIC is a rather pedantic distinction IMHO :)

By the way, someone has already developped a full custom ASIC for Bitcoin mining: http://3.14.by/en/read/bitfury-bitcoin-mining-chip

Will these things be good for anything else once faster/better miners are introduced?

what kind of faster miners are you referring to? ASIC is the best available technology

The chip technology is an older process using 110nm technology. There are already several companies working on 40nm technology which will be more power efficient and be significantly faster per chip.

110nm is old news. Butterfly labs do 65 nm and KnC does 28 nm. I guess 110 was a test run really, or they lacked access to plants that could do smaller chips.

The avalon chips the author is describing in the article are 110nm I believe.

They will keep making money, just slower.

Will the amount of money they make pay for the power they consume?

No, but somebody who has free power may be willing to buy your obsolete used ASICs cheap.

therein lies the rub.

ASICs are very power efficient.

Plus the system is self-correcting. Mining becomes unprofitable, people quit, complexity drops enough, mining becomes profitable.

No, they are specifically built for mining bitcoin.

Miners of this type can be pointed to other cryptocurrencies that use the same proof-of-work function.

You would expect anyone else in the world to avoid compatible proof-of-work functions for exactly that reason.

password hash cracking, perhaps? If they're using the same SHA hash.

I for one, look forward to all the cheap 6990s that are about to hit eBay.

And Butterfly Labs still hasn't shipped their orders from Day 1. There will be an awful lot of people who will never see a return on their investment from this venture.

Would you ship them if you could make money by simply using them yourself? I'm a bit skeptical of this whole situation.

It looks like they're shipping orders up to Oct 2012 now: https://forums.butterflylabs.com/blogs/bfl_jody/

Or am I reading that wrong?

Doesn't the reliability of bitcoin stem from it being very difficult for any one entity to control over 50% of the computing power in the network? News like this makes it seem like a state sponsored actor could do it without too much trouble.

It's easy to get over 50% of the hashing power: just buy it. Bitcoin is especially vulnerable during technology transitions; e.g. in July 2010 a single Radeon GPU was more powerful than the entire Bitcoin network so you could have performed a 51% attack for $300 plus some epic hacking.

Right, but of course Satoshi's argument is that it would be far more valuable for you to have just used that Radeon to mine ;)

> Right, but of course Satoshi's argument is that it would be far more valuable for you to have just used that Radeon to mine

Value depends on personal desire... Not all people want the same things.

That's why we use money to exchange goods and services ;)

The flood of newly minted BTC will likely cause exchange rates to crater. Hopefully the increase in available BTC will stabilize prices as individuals/small groups will have to expend more capital to influence prices in their favor.

The next six months are probably going to be a little crazy, hopefully it'll drive BTC out of the hands of speculators and into the hands of actual users of the currency.

The creation rate of BTC is fixed every 10 minutes. More ASIC power does not equal more BTC.

Not quite. The difficulty is set to average that a new block is found approximately every 10minutes. However, if lots of capacity is added all at once, the difficulty might not change for several days, and blocks will be generated at a faster rate during that time, as it is only updated every 2016 blocks.

https://en.bitcoin.it/wiki/Difficulty

Somewhat splitting hairs... it's adjusted to 10 minute blocks, and there's also a huge variability in how frequently they are discovered. Even if the average was 10 min, there are blocks found within the same minute, or with 30+ minute gaps.

Variability wouldn't matter if the average over time was alright. But it's not just spitting hairs: BTC generation is significantly ahead of schedule. https://news.ycombinator.com/item?id=5780861

Creation is indeed fixed and constant, but the supply of BTC within the exchanges isn't. The new owners of these ASIC boxes will find themselves in an excess of supply and they'll quite happily flood the market with this increased supply, or they'll spend the coins at merchants.

Contrastingly other miners with inferior machines will find themselves behind the curve with a smaller source of income.

I doubt it. After the a few hundred (thousand?) blocks are mined, the mining difficulty algorithm will kick in, making mining just as difficult as it was before, relative to the new hash rate. It will mean gpu miners won't make anything at all, but the overall volume of new coins won't increase.

Further reading: https://en.bitcoin.it/wiki/Difficulty

There isn't really a "flood" of newly minted BTC.

The system performs closed loop control keeping the supply relatively constant.

Supply as a whole is constant, but supply between exchanges and people isn't constant.

Absolutely, but what makes you think that the people with increased hash power will have less tight handed behavior than the others? Why not the opposite?

Would a quantum computer be able to solve these kind of problems instantly?

The difficulty would rise to compensate for the advent of quantum computing. The danger of quantum computing in bitcoin is holding the 51%+ computing power and being able to forge transactions in the 'books'.

Source: http://bitcoin.stackexchange.com/questions/6062/what-effects...

Hashing isn't affected by quantum computing. Quantum computing would only affect Bitcoin if the public-key aspect is vulnerable to quantum computing.

Reply on news.ycombinator.com