One of the big complaints about Bitcoin -- and cryptocurrencies in general -- is that they're using "tons" of power and not really accomplishing any "useful" purpose. While it's certainly true that all of the ASICs out there hashing away to secure the Bitcoin network (and mine coins in the process) use power, it's important to put things in perspective. Let's start with a rough estimate of the power used by all the systems connected to the Bitcoin network.
At present, the total hash rate of the Bitcoin network is around 240,000 TH/s, though of course that changes on almost a daily basis -- you can check the current approximate hash rate any time at BitcoinWisdom (or any number of other sites). The most efficient Bitcoin ASICs right now can do around 3 TH/s while drawing 2000 W (give or take), while upcoming ASICs may be as much as two to four times as efficient (e.g. around 3 TH/s while drawing only 600 W). Obviously not every ASIC currently hashing on the Bitcoin network is going to be the most efficient option (side note: I finally shut down my AntMiner S2 ASICs as their hash rate to power is no longer profitable), but let's just estimate that most ASICs today are averaging a 1:1.5 ratio of TH to power.
What that means is that our 240,000 TH/s of hash rate is using 360,000 kW. (By way of comparison, if every current ASIC was a KnC Neptune doing 3 TH per 2000 W, the Bitcoin network would only use 160,000 kW.) 360,000 kW becomes 360 MW, and 24 hours per day means the Bitcoin network is using around 8640 MWh (Megawatt Hours) each day, and 259,200 MWh per month. So how does that compare with the power used for other tasks?
According to Wikipedia, the total power consumed in the US per month was 2,183 TWh, or 2,183,000,000 MWh. That means the total power used by the entire Bitcoin network is approximately 0.012% of the US energy use. But Bitcoin isn't just a US phenomenon, so we really need to look at the entire world. In 2008 the worldwide approximate power use was 143,851,000,000 MWh, or 11,823,000,000 MWh per month. It's likely power use has increased since then, but let's just stick with that number for now. That means the Bitcoin network "waste" of power accounts for a whopping 0.0022% of all energy consumed in the world.
Let me put that figure in different terms. The average power draw of a US household is 10,837 kWh per year, or 29.7 kWh per day. That's the same as a continual power draw of around 1250W from all the lights, appliances, computers, etc. in your home. 0.0022% of the average power used by an American home would equal 0.0275 W. Or in other words, shutting down all Bitcoin related hardware in the world would be like the average American home cutting their power use by $0.00008 per day.
Now, it's entirely possible I screwed up on the math somewhere. Maybe I'm even off by a factor of 1000, but I'm pretty sure that's not the case. I've checked things multiple times and I think I've got it reasonably close, understanding that this is merely an estimate and I could easily be off by a factor of 2-4X on a few guesses (e.g. maybe the average efficiency of ASICs is much lower than my estimate). Still, if you spot an error, by all means let me know.
The bottom line is that the amount of power being used globally for the Bitcoin network is pretty tiny as a percentage. When you think of all the other things out there consuming power -- lights on empty roads and parking lots, computers that sit idle at large corporations, etc. -- there are far worse ways of using power. Inherently, people and businesses consume power because they find it to be a good use of their money -- cost vs. benefit. There's no need to try and halt the use of power by cryptocurrencies as they'll eventually reach equilibrium on their own.
Bitcoin hashing might be consuming 360 MW (which is still about one fourth of the infamous 1.21 Gigawatts number from Back to the Future), but how much power is consumed by all of our financial institutions? I'm absolutely sure it's far more than 360 MW, so by that token the Bitcoin network is actually a much more efficient way of doing things.
Lots of nerdy stuff, mostly focused on graphics cards, cryptocurrency, and related technologies.
Wednesday, September 24, 2014
Tuesday, September 23, 2014
PayPal (Finally) Offers Bitcoin Support
So why is PayPal willing to work with Bitcoin now and not last year, or even two or three years ago? I think the inability to roll back purchases made via Bitcoin posed a serious risk, especially for fraud. Just imagine complaints like, "Hey, I bought this $2000 product with Bitcoin and the merchant never shipped it!" PayPal would have been forced to either refund the money from its own pockets, or perhaps worse the user would end up being scammed. With companies like Coinbase and BitPay now providing services that help mitigate some of these risks, plus successful adoptions of Bitcoin by many other resellers, it looks like PayPal is finally ready to hop on the Bitcoin bandwagon. All I can say is that it's about time!
The support for Bitcoin will come via integration into the PayPal Payments Hub, and unfortunately it will only be supported for merchants in North America for the time being. There are other qualifications to using Bitcoin with PayPal as well, as the blog notes: "To be clear, today’s news does not mean that PayPal has added Bitcoin as a currency in our digital wallet or that Bitcoin payments will be processed on our secure payments platform. PayPal has always embraced innovation, but always in ways that make payments safer and more reliable for our customers. Our approach to Bitcoin is no different. That’s why we’re proceeding gradually, supporting Bitcoin in some ways today and holding off on other ways until we see how things develop."
Of note is that this comes at a time when the mining phase of Bitcoins and other virtual currencies has largely moved beyond GPUs and onto dedicated SHA256 and Scrypt ASICs. That's good news for gamers and graphics gurus, though I suspect the GPU vendors may not sell as many new GPUs – and certainly not at such inflated prices – as they did last year. Then again, I really, really want to see what a GeForce GTX 980 (or GeForce GTX 970) can do with mining – not that I'm willing to spend $549 to start mining with one, but given the improvements in efficiency it could actually do okay at mining Cryptonight, X11, X13, etc. Of course I suspect the GPU vendors will also have fewer RMAs over the coming year – I know between myself and a friend, we've had at least two R9 290X GPUs and four HD 7950 cards go out due to the strains of 24/7 cryptocurrency mining (and overclocking certainly didn't help).
Wrapping things up, PayPal also notes that PayPal has worked with merchants selling cryptocurrency mining hardware but refuses to support pre-orders, which is a stance I wholly support having now been more or less burned by Alpha Technologies. And while the announcement today is more of a baby step as opposed to wholesale acceptance of Bitcoin, the closing comments suggest that this is merely the tip of the iceberg: "PayPal is excited about all the innovations taking place in payments these days. More choices in how people create value, share it, buy, sell and trade it – that’s exactly what PayPal is all about. And we believe Bitcoin offers unique opportunities as more people and businesses experiment with it. We are excited to work with businesses and business models that allow us to offer new experiences and the trusted service our customers expect. We hope to do more with Bitcoin as its ecosystem continues to evolve."
I've now successfully used Bitcoin to purchase quite a few goods, from graphics cards on Newegg.com to the games in the latest Humble Bundle, and more and more places are beginning to get onboard. For example TigerDirect and Overstock now accept Bitcoin as a viable method of payment, and you can use services like Gyft or eGifter to purchase a gift card with Bitcoins that can be used at numerous other locations (including Amazon.com). They might be a bit late to the part, but it looks like PayPal has decided to join the club.
While there are still naysayers when it comes to Bitcoins and cryptocurrencies in general, this is great news for Bitcoin proponents. Integration with PayPal literally opens the doors for thousands of small shops to easily begin accepting Bitcoin. It's one small step for PayPal, one giant leap for Bitcoin.
Monday, September 22, 2014
Welcome to the End of GPU Mining
It's been a long, slow decline, but outside of a major jump in cryptocurrency prices we have finally reached the end of the road for GPU mining. That's not to say that people won't continue to mine with GPUs, but based on the cost of electricity and the current price of Bitcoin combined with the alt-coin exchange rates...well, you get the point. GPU mining is "dead". Let's quickly run some figures to give you an idea of where we stand.
The best-case mining opportunity right now (as far as I can tell -- there are so many coins that it's possible I'm missing some "secret" coin that's highly profitable) would be GPU mining of Monero, one of the Cryptonight coins. With the current difficulty, block reward, exchange rate, and price of BTC, a rig with three R9 280X GPUs might mine around 1.5 XMR per day, which would equal about $1.90. That same rig will draw about 450W of power, and at $0.10 per kWh you'd spend $1.08 per day in electricity. Factor in wear and tear on the GPUs and the net profits of $0.80 per day probably isn't worth your time, but that's about the best option for GPUs right now.
Okay, that's XMR and Cryptonight, but what about some of the other proof of work algorithms that can't be run on an ASIC? We have X11, X13, X15, Keccak, and maybe a few others as potential PoW options. There are a few others as well, but let's just stick with those for now.
X11 was made famous with Darkcoin, which is currently averaging a difficulty of around 3300 with a block reward of 4 DRK being typical. If your hash rate is 12000 KH/s for X11 (which is about right for three 280X GPUs), and your power draw is only 450W, you'd still be in the red. You might mine 0.3 DRK per day, worth about $0.90, with a power cost of $1.08. Even R9 290X wouldn't be much better: 16 MH/s for three 290X GPUs and at a power use of 500W you'd basically break even. CANN is about the same right now, so is URO, and basically we're looking at best-case breaking even or making pennies per day per GPU.
Actually, your best bet is often to just go with leasing your hash rates on a service like NiceHash -- let others do the work of figuring out which new alt-coins to mine, let them take the risk, and you just get paid directly in BTC. Or go with a multi-pool like BlackcoinPool, MultiPool, WafflePool, etc. The current best rates for X11 are around 0.0003 BTC per MH per day, so three R9 290X could make about $1.90 per day, with a power cost of $1.25 (give or take). That's "profitable", but not enough that I'd seriously invest time and effort into it. So X11 is out, unless prices go up, difficulty goes down, or hash rates can be increased without using more power.
X13 and X15 really aren't much different in concept from X11, just with lower hash rates thanks to the additional hashing algorithms. X13 you might make a bit more per MH (currently looking like 0.00037 BTC/MH/day, give or take), but with lower hash rates in general it's less profitable (or more unprofitable). X15 is paying less than X11 and X13 with even lower hash rates, so obviously that's out as well.
If you're after really high hash rates, you could look to Keccak, but just because it has high hash rates (450MH for a 290X?) doesn't mean it's profitable. The going rate is about 0.0000023 BTC/MH/Day, or $1.25 per day for 3x290X. Nist5 is another "faster" hashing PoW, but it's only about three times the hash rate of X11 and it's paying about 10% of X11, so that's a dead end as well.
The net result is that across the current list of coins and PoW algorithms, there's really nothing out there that can even hit a 2:1 ratio for value vs. power. The best I can currently find is Monero, which gives about a 1.7:1 ratio of income vs. electricity. If you were to use a $300 R9 280X GPU, not accounting for the rest of the PC, you would need to mine Monero for over two years straight just to break even. And let me tell you, my personal experience is that running a GPU 24/7 at 100% load (which is what you do with cryptocurrencies) could very well kill the GPU before two years pass -- or at least the GPU fans will likely need to be replaced.
If you're wondering why I've become so pro-ASIC (or pro-Hashlet), this is the answer. All of my GPUs are now powered down, and I for one welcome the blessed silence. Maybe I'll fire them back up to help heat my house as it gets colder, or maybe prices will shoot up again and make GPU mining profitable. Until one of those things happen -- or something else comes along to make me want to mine with GPUs -- I'm going to be investing my time and energy elsewhere. Anyone interested in buying some GPUs?
The best-case mining opportunity right now (as far as I can tell -- there are so many coins that it's possible I'm missing some "secret" coin that's highly profitable) would be GPU mining of Monero, one of the Cryptonight coins. With the current difficulty, block reward, exchange rate, and price of BTC, a rig with three R9 280X GPUs might mine around 1.5 XMR per day, which would equal about $1.90. That same rig will draw about 450W of power, and at $0.10 per kWh you'd spend $1.08 per day in electricity. Factor in wear and tear on the GPUs and the net profits of $0.80 per day probably isn't worth your time, but that's about the best option for GPUs right now.
Okay, that's XMR and Cryptonight, but what about some of the other proof of work algorithms that can't be run on an ASIC? We have X11, X13, X15, Keccak, and maybe a few others as potential PoW options. There are a few others as well, but let's just stick with those for now.
X11 was made famous with Darkcoin, which is currently averaging a difficulty of around 3300 with a block reward of 4 DRK being typical. If your hash rate is 12000 KH/s for X11 (which is about right for three 280X GPUs), and your power draw is only 450W, you'd still be in the red. You might mine 0.3 DRK per day, worth about $0.90, with a power cost of $1.08. Even R9 290X wouldn't be much better: 16 MH/s for three 290X GPUs and at a power use of 500W you'd basically break even. CANN is about the same right now, so is URO, and basically we're looking at best-case breaking even or making pennies per day per GPU.
Actually, your best bet is often to just go with leasing your hash rates on a service like NiceHash -- let others do the work of figuring out which new alt-coins to mine, let them take the risk, and you just get paid directly in BTC. Or go with a multi-pool like BlackcoinPool, MultiPool, WafflePool, etc. The current best rates for X11 are around 0.0003 BTC per MH per day, so three R9 290X could make about $1.90 per day, with a power cost of $1.25 (give or take). That's "profitable", but not enough that I'd seriously invest time and effort into it. So X11 is out, unless prices go up, difficulty goes down, or hash rates can be increased without using more power.
X13 and X15 really aren't much different in concept from X11, just with lower hash rates thanks to the additional hashing algorithms. X13 you might make a bit more per MH (currently looking like 0.00037 BTC/MH/day, give or take), but with lower hash rates in general it's less profitable (or more unprofitable). X15 is paying less than X11 and X13 with even lower hash rates, so obviously that's out as well.
If you're after really high hash rates, you could look to Keccak, but just because it has high hash rates (450MH for a 290X?) doesn't mean it's profitable. The going rate is about 0.0000023 BTC/MH/Day, or $1.25 per day for 3x290X. Nist5 is another "faster" hashing PoW, but it's only about three times the hash rate of X11 and it's paying about 10% of X11, so that's a dead end as well.
The net result is that across the current list of coins and PoW algorithms, there's really nothing out there that can even hit a 2:1 ratio for value vs. power. The best I can currently find is Monero, which gives about a 1.7:1 ratio of income vs. electricity. If you were to use a $300 R9 280X GPU, not accounting for the rest of the PC, you would need to mine Monero for over two years straight just to break even. And let me tell you, my personal experience is that running a GPU 24/7 at 100% load (which is what you do with cryptocurrencies) could very well kill the GPU before two years pass -- or at least the GPU fans will likely need to be replaced.
If you're wondering why I've become so pro-ASIC (or pro-Hashlet), this is the answer. All of my GPUs are now powered down, and I for one welcome the blessed silence. Maybe I'll fire them back up to help heat my house as it gets colder, or maybe prices will shoot up again and make GPU mining profitable. Until one of those things happen -- or something else comes along to make me want to mine with GPUs -- I'm going to be investing my time and energy elsewhere. Anyone interested in buying some GPUs?
Saturday, September 20, 2014
Bitcoin Prices: Deja Vu All Over Again
Newcomers to the world of Bitcoin and cryptocurrencies might be wetting their pants right about now, depending on when they decided to get on the bus. We've gone from a high of around $660 in July to the current price of just over $400, with dips as low as $375 (give or take). The early hype stages of Bitcoin are now clearly over and it's time for things to slide into oblivion... or at least that's the pessimistic view of things. But you know what's interesting about pessimists? They've been predicting the demise of Bitcoin for well over three years now, and so far they've been wrong every time.
Yes, we've had ups and downs, some bigger than others, and there have been a variety of reasons for the spikes and valleys. Major accounts have been hacked and dumped, exchanges and other services have evaporated into thin air with large numbers of Bitcoin, China has jumped on -- and then been pushed off -- Bitcoin. But even without some of those events, Bitcoin like all markets is going to have upward and downward trends. And right now, we're clearly in a downward trend, but here's the question: how low will we go, and when the inevitable rebound occurs, how high will we climb?
2013 was, by all accounts, a very exciting year for Bitcoin. It took a while, but we finally eclipsed the previous high of June 2011... and kept on going! Around early April, Bitcoin reached a then unheard of high of around $260, at which point there was a major sell-off and some other stuff like the MtGox fiasco that caused a crash back to the $50 range. Then we went back to $170, down to $80, up to $130, and then there was a downward trend to about $66 again. The naysayers were out in full force, as you might guess, and then things started to pick up around August.
Bitcoin went up to $130 or so again, and basically hovered around that mark for a bit. There was a big dump to $85 in late September (I think that might have been an exchange getting hacked or something similar), and then we started to jump up in prices. The first plateau was back to $200, and I'll admit I cashed out of quite a few Bitcoins at that point. Over the next month I just had to shake my head as we repeatedly set new highs, eventually topping at over $1150.
After such a rapid climb, the fall-off was almost as dramatic, but it was far more seesaw in nature. Slowly but surely we've been heading down again, but we're still at more than three times the prices of last September.
Now, I'm not going to guarantee anything with Bitcoin right now, especially not on any short-term time frame, but what I will say is that this downward slump is nothing new. I suspect we'll even go as low as $300 or even $200, though that's a gamble if you're going to sell now and try to buy back in, but mark my words: Bitcoin is going to go back up again.
My take is that the holiday season is about to kick into full swing, and when people start buying lots of gifts and other products, companies start looking like they're doing well and stocks go up. Bitcoin most definitely isn't a stock, but there are many investors that play the stock market and they're now playing Bitcoin, so it's often treated like a stock. That being the case, the end of this year and the start of 2015 could be as exciting as last year, perhaps even more so.
Personally, I'm hanging onto all the Bitcoins I can get my grubby little hands on right now, and my Zen Hashlets are helping me accumulate more at a nice steady rate. Long-term, I think the next major spike in Bitcoin prices will eclipse $2000, and possibly go as high as $5000, and then we'll see the usual collapse again as prices consolidate around a new high that will very likely be over the $1000 mark. If you sell now, you might become yet another one of those "weak hands being shaken out", though as long as you're not selling at a loss I suppose that's fine. Just remember: the real winners are the people that successfully play the long ball.
Come talk to me again in 2015 and let's see how my prediction pans out. Again, no guarantees, but I think we're going to see a major upswing some time between now and January 2015. That's where I'd place my bets at least, if I were a betting man. And since Bitcoin is almost like gambling, maybe in this sense I am a betting man.
Yes, we've had ups and downs, some bigger than others, and there have been a variety of reasons for the spikes and valleys. Major accounts have been hacked and dumped, exchanges and other services have evaporated into thin air with large numbers of Bitcoin, China has jumped on -- and then been pushed off -- Bitcoin. But even without some of those events, Bitcoin like all markets is going to have upward and downward trends. And right now, we're clearly in a downward trend, but here's the question: how low will we go, and when the inevitable rebound occurs, how high will we climb?
2013 was, by all accounts, a very exciting year for Bitcoin. It took a while, but we finally eclipsed the previous high of June 2011... and kept on going! Around early April, Bitcoin reached a then unheard of high of around $260, at which point there was a major sell-off and some other stuff like the MtGox fiasco that caused a crash back to the $50 range. Then we went back to $170, down to $80, up to $130, and then there was a downward trend to about $66 again. The naysayers were out in full force, as you might guess, and then things started to pick up around August.
Bitcoin went up to $130 or so again, and basically hovered around that mark for a bit. There was a big dump to $85 in late September (I think that might have been an exchange getting hacked or something similar), and then we started to jump up in prices. The first plateau was back to $200, and I'll admit I cashed out of quite a few Bitcoins at that point. Over the next month I just had to shake my head as we repeatedly set new highs, eventually topping at over $1150.
After such a rapid climb, the fall-off was almost as dramatic, but it was far more seesaw in nature. Slowly but surely we've been heading down again, but we're still at more than three times the prices of last September.
Now, I'm not going to guarantee anything with Bitcoin right now, especially not on any short-term time frame, but what I will say is that this downward slump is nothing new. I suspect we'll even go as low as $300 or even $200, though that's a gamble if you're going to sell now and try to buy back in, but mark my words: Bitcoin is going to go back up again.
My take is that the holiday season is about to kick into full swing, and when people start buying lots of gifts and other products, companies start looking like they're doing well and stocks go up. Bitcoin most definitely isn't a stock, but there are many investors that play the stock market and they're now playing Bitcoin, so it's often treated like a stock. That being the case, the end of this year and the start of 2015 could be as exciting as last year, perhaps even more so.
Personally, I'm hanging onto all the Bitcoins I can get my grubby little hands on right now, and my Zen Hashlets are helping me accumulate more at a nice steady rate. Long-term, I think the next major spike in Bitcoin prices will eclipse $2000, and possibly go as high as $5000, and then we'll see the usual collapse again as prices consolidate around a new high that will very likely be over the $1000 mark. If you sell now, you might become yet another one of those "weak hands being shaken out", though as long as you're not selling at a loss I suppose that's fine. Just remember: the real winners are the people that successfully play the long ball.
Come talk to me again in 2015 and let's see how my prediction pans out. Again, no guarantees, but I think we're going to see a major upswing some time between now and January 2015. That's where I'd place my bets at least, if I were a betting man. And since Bitcoin is almost like gambling, maybe in this sense I am a betting man.
Wednesday, September 17, 2014
BitMain AntMiner S4 - 2TH Coming Soon, ROI Estimates
First, we need to know roughly how much power the AntMiner S4 will use. The previous generation AntMiner S3 is a 28nm chip, but it's only moderately efficient -- 441GH at 340W, or 0.77 W/GH. The AntMiner S4 might simply be the equivalent of five of S3 blades slapped into a single chassis, clocked lower and running at a lower voltage to improve efficiency. If that's the case, we might see 0.6 or even 0.5 W/GH.
The other item we don't know is the price, and given the above speculation I'd say it's likely we'll see the price in the neighborhood of $2000. Anything higher than that is obviously a bad investment -- you can buy a 10GH Hashlet Genesis for $8, basically, so that's the starting point I'm working off. Maybe BitMain will be kind to their customers and drop the price as low as $1500, but that's about the minimum I'd expect to see. We also need to know if it includes a power supply or if you need to provide your own, but the image at least suggests it will be a fully "plug and play" affair.
ROI Estimates and Early Forecasts
Considering we have a reasonable idea of the hardware and price, I can at least do a quick estimate of ROI potential. Let's start with the worst-case estimate: you pay $2000, it uses 1000W, and BTC difficulty increases an average of 20% every two weeks; power cost will be a reasonable $0.10 per kWh. We'll guess that you get the hardware in 14 days and start mining immediately. If this "worst-case" happens to be anywhere close to reality, you'd basically spend $2000 and your new ASIC would be too power hungry to keep running in about a third of a year (115 days, give or take). Assuming BTC price stays where it's at, the forecast would be pretty bleak: you'd lose around $1550 (with free shipping no less)!Okay, that's too grim a proposal, as I think 20% difficulty increase is way too high. But if we take the same core values for price, power draw, and start date we can figure out what the average rate of increase needs to be for you to break even... and it's no less grim! If BTC difficulty only goes up 5.29% each cycle, a 2000GH ASIC for $2000 starting mining in two weeks will basically break even, but that won't happen until early 2016.
I don't think anyone would be willing to buy the S4 at that price, so let's try again and be more optimistic. Let's guess that they can get power draw down to 750W instead of 1000W, and the price is only $1500. Using our more than generous estimates from above (5.29% difficulty increase) you would hit ROI in about six months. A less optimistic forecast of difficulty going up 7.9% on average would leave you with a break even investment in one year.
That's still not looking very good, so let's try once more. What if the AntMiner S4 only costs $1000 and uses 500W? Now we're talking: at 5.29% you'd make ROI in just three months (~95 days), and long-term you could end up earning $1665 off of your $1000, though it would still take the better part of two years. 60% (or more) annual interest would be pretty awesome, so at these (nearly fantasy land) estimates the S4 would be worth buying. Of course a more likely 10% average increase in Bitcoin difficulty would drop you to hitting ROI in four months (~130 days) and after a year your total profit would only be $343.
If you can't tell, we're getting awfully close to the end of the road for Bitcoin ASIC mining investments. If you have one already and it's still profitable, you keep running it, but buying new hardware is looking sketchy at best. The days of purchasing a $500 GPU and having it pay for itself in just a month or so of mining are long past, so short of a major spike in BTC prices we're likely hitting a BTC difficulty plateau that will last a long time and only inch upward slowly as new process technology is developed.
BitFury Group's Upcoming ASIC: More Efficient SHA256 Hashing
Earlier this week, the BitFury Group issued a technology roadmap update for their ASICs. What's interesting about BitFury is that unlike many of their competitors, they've apparently put a lot more work into designing a custom ASIC. I'll get to what this means in a moment, but first we need to take a step back and discuss general microprocessor design principles.
Designing a CPU, GPU, SoC, ASIC, etc. can be done in many ways, but from a high level there are two general approaches. One is to use machine algorithms to optimize and lay out the transistors, and the other is to basically design the logic circuits and do the layout "by hand". There are pros and cons to either approach, of course.
The machine algorithms can do a great job of testing and validating the design and basically get you up and running a lot faster than if you were to have a human (many humans) perform the same work. What's more, there are many companies that now sell functional blocks of compute logic, so you can integrate these functional blocks into your chip a lot easier if you let the machines do the legwork. The drawbacks to machine layouts are that they typically use more area and they tend to be less power efficient. These aren't necessarily inviolate rules, but that's the basic idea.
Doing the layout by hand is basically the reverse of the above: it can take much longer to complete the design, validation, testing, etc. However, a human can generally see the big picture better than a machine and so they can optimize better for die area as well as power. This can in turn lead to potentially higher performance, which can be very important for high performance (or low power) microprocessors.
The above is a low-level discussion of processor design, but one of the interesting ideas is the use of ready made functional blocks. If you've ever wondered why it took a while to see the first SHA256 (Bitcoin) ASICs and then suddenly there was an explosion of competing designs from several companies, it's because once there was a tested and validated solution available, many other companies were able to license/buy the basic design and then just place more chips on a board to improve performance.
There are still multiple Bitcoin ASIC designs of course. The earliest ASICs were built on 90nm or even 130nm process technology (because it was cheap, mature, readily available, and easier to use), but as the competition heated up things shifted to newer and smaller process nodes. Today, the fastest and most efficient ASICs are manufactured on 28nm process technology, and 20nm designs will probably come out within the next year (the 20nm fabrication facilities are busy making things like Apple's A8 and the new Qualcomm Snapdragon cores, so they would cost a lot more to use). However, even 55nm ASICs can still be efficient enough to earn a profit -- or at least, the power cost of running them is lower than the value of BTC they generate.
BitFury is a prime example of this last case, as up until now they have been using 55nm process technology. The key to staying competitive even with an older process node is that BitFury uses their own custom logic (i.e. it's not licensed from another company), and they apparently put a bit more effort into optimizing for power and efficiency. Or more likely, they run at lower clocks and they're not really performance competitive right now -- the current BitFury ASICs can hit 3.5 TH/s at 2800W (give or take), but the power use is likely the limiting factor.
The latest announcement is basically BitFury Group saying that their 28nm custom logic ASIC is nearly ready. With the smaller process node and additional time spent optimizing for power efficiency, BitFury is claiming that they will have ASICs capable of running at 0.2 J/GH (essentially 0.2 W/GH) by the end of 2014, most likely late December. They're also working on an even more efficient design that will use 0.1 J/GH (W/GH, assuming 0.1 J/s) in mid-2015. That doesn't really tell us a lot in a vacuum, though, so let's compare those power numbers with some existing ASICs.
BitFury's own 3500BF I mentioned above delivers 3500 GH/s at 2800W, so it's doing about 0.8 W/GH. It also uses 1320 BF864C55 chips, which can run at a voltage range of 0.5V to 1.2V depending on your desired efficiency, with 0.5 J/GH being the maximum efficiency while 3.8 GH/s is the maximum performance -- but you have to choose one or the other. (Ever wonder why you can overclock ASICs? It's because you're just trading efficiency for higher performance, so if you have an older ASIC that's pulling 420W at the wall and you drop the clocks 10%, you'll likely end up improving efficiency by more than 10%.)
The KnC Neptune is currently targeting 3500 GH/s at 1950W, so it's slightly more efficient than the 3500BF (0.56 W/GH), but it's already on 28nm. Butterfly Labs' Monarch is more like existing chips, as it's capable of 700 GH/s at 490W (0.7 W/GH). Bitmain has their AntMiner S3 that's also around 0.78 W/GH, though the Antminer S4 is "coming soon". Looking at a list of other ASICs, those figures are pretty similar to the other "state of the art" designs.
Since I've been on a Hashlet's kick, I might as well toss out the Hashlet Genesis as well. GAW isn't saying how much power the Genesis actually uses, but the cost to run it (hosting included) is $0.02 per 10 GH. Doing the math at $0.10 per kWh, that would mean GAW is basically charging you at a rate equivalent to roughly 0.83 W/GH.
Basically, if we look at most of the currently shipping Bitcoin ASICs, the best you might get out of them is 0.5 W/GH, so BitFury Group is claiming they will more than double the hashing efficiency, and by the middle of next year they'll double it again. It's not just about efficiency of course -- the initial price will also largely determine whether or not a new ASIC is worth buying, so keep that in mind.
As I've said in the past, the real money makers in the whole Bitcoin Gold Rush are the people selling the mining hardware -- or other services as the case may be. You can't expect them to give the stuff away, obviously, but they're taking a healthy profit in most cases and hitting ROI is sometimes difficult (especially when the manufacturers mine with the hardware for a month or two before shipping to customers). Hopefully the 28nm BitFury parts get to end customers sooner rather than later, as we're getting close to the point where many of the current ASICs are going to have to be retired. Anything worse than about 3 W/GH is now breaking even on power, but really you'd want to be below 1.5 W/GH to keep mining viable -- and if you pay more for electricity (like $0.30 per kWh), you'd be breaking even at just 1 W/GH!
Anyway, for those thinking the current levels of efficiency were the end of the road for Bitcoin ASICs, there's still plenty of room left for optimizations. The first wave is now over (and probably the second and third as well), and the focus is now on refining designs rather than just getting them out the door. It's going to be interesting to see what sort of pricing we get on the next generation of ASICs, but we're still a few months away it looks like.
Designing a CPU, GPU, SoC, ASIC, etc. can be done in many ways, but from a high level there are two general approaches. One is to use machine algorithms to optimize and lay out the transistors, and the other is to basically design the logic circuits and do the layout "by hand". There are pros and cons to either approach, of course.
The machine algorithms can do a great job of testing and validating the design and basically get you up and running a lot faster than if you were to have a human (many humans) perform the same work. What's more, there are many companies that now sell functional blocks of compute logic, so you can integrate these functional blocks into your chip a lot easier if you let the machines do the legwork. The drawbacks to machine layouts are that they typically use more area and they tend to be less power efficient. These aren't necessarily inviolate rules, but that's the basic idea.
Doing the layout by hand is basically the reverse of the above: it can take much longer to complete the design, validation, testing, etc. However, a human can generally see the big picture better than a machine and so they can optimize better for die area as well as power. This can in turn lead to potentially higher performance, which can be very important for high performance (or low power) microprocessors.
The above is a low-level discussion of processor design, but one of the interesting ideas is the use of ready made functional blocks. If you've ever wondered why it took a while to see the first SHA256 (Bitcoin) ASICs and then suddenly there was an explosion of competing designs from several companies, it's because once there was a tested and validated solution available, many other companies were able to license/buy the basic design and then just place more chips on a board to improve performance.
There are still multiple Bitcoin ASIC designs of course. The earliest ASICs were built on 90nm or even 130nm process technology (because it was cheap, mature, readily available, and easier to use), but as the competition heated up things shifted to newer and smaller process nodes. Today, the fastest and most efficient ASICs are manufactured on 28nm process technology, and 20nm designs will probably come out within the next year (the 20nm fabrication facilities are busy making things like Apple's A8 and the new Qualcomm Snapdragon cores, so they would cost a lot more to use). However, even 55nm ASICs can still be efficient enough to earn a profit -- or at least, the power cost of running them is lower than the value of BTC they generate.
BitFury is a prime example of this last case, as up until now they have been using 55nm process technology. The key to staying competitive even with an older process node is that BitFury uses their own custom logic (i.e. it's not licensed from another company), and they apparently put a bit more effort into optimizing for power and efficiency. Or more likely, they run at lower clocks and they're not really performance competitive right now -- the current BitFury ASICs can hit 3.5 TH/s at 2800W (give or take), but the power use is likely the limiting factor.
The latest announcement is basically BitFury Group saying that their 28nm custom logic ASIC is nearly ready. With the smaller process node and additional time spent optimizing for power efficiency, BitFury is claiming that they will have ASICs capable of running at 0.2 J/GH (essentially 0.2 W/GH) by the end of 2014, most likely late December. They're also working on an even more efficient design that will use 0.1 J/GH (W/GH, assuming 0.1 J/s) in mid-2015. That doesn't really tell us a lot in a vacuum, though, so let's compare those power numbers with some existing ASICs.
BitFury's own 3500BF I mentioned above delivers 3500 GH/s at 2800W, so it's doing about 0.8 W/GH. It also uses 1320 BF864C55 chips, which can run at a voltage range of 0.5V to 1.2V depending on your desired efficiency, with 0.5 J/GH being the maximum efficiency while 3.8 GH/s is the maximum performance -- but you have to choose one or the other. (Ever wonder why you can overclock ASICs? It's because you're just trading efficiency for higher performance, so if you have an older ASIC that's pulling 420W at the wall and you drop the clocks 10%, you'll likely end up improving efficiency by more than 10%.)
The KnC Neptune is currently targeting 3500 GH/s at 1950W, so it's slightly more efficient than the 3500BF (0.56 W/GH), but it's already on 28nm. Butterfly Labs' Monarch is more like existing chips, as it's capable of 700 GH/s at 490W (0.7 W/GH). Bitmain has their AntMiner S3 that's also around 0.78 W/GH, though the Antminer S4 is "coming soon". Looking at a list of other ASICs, those figures are pretty similar to the other "state of the art" designs.
Since I've been on a Hashlet's kick, I might as well toss out the Hashlet Genesis as well. GAW isn't saying how much power the Genesis actually uses, but the cost to run it (hosting included) is $0.02 per 10 GH. Doing the math at $0.10 per kWh, that would mean GAW is basically charging you at a rate equivalent to roughly 0.83 W/GH.
Basically, if we look at most of the currently shipping Bitcoin ASICs, the best you might get out of them is 0.5 W/GH, so BitFury Group is claiming they will more than double the hashing efficiency, and by the middle of next year they'll double it again. It's not just about efficiency of course -- the initial price will also largely determine whether or not a new ASIC is worth buying, so keep that in mind.
As I've said in the past, the real money makers in the whole Bitcoin Gold Rush are the people selling the mining hardware -- or other services as the case may be. You can't expect them to give the stuff away, obviously, but they're taking a healthy profit in most cases and hitting ROI is sometimes difficult (especially when the manufacturers mine with the hardware for a month or two before shipping to customers). Hopefully the 28nm BitFury parts get to end customers sooner rather than later, as we're getting close to the point where many of the current ASICs are going to have to be retired. Anything worse than about 3 W/GH is now breaking even on power, but really you'd want to be below 1.5 W/GH to keep mining viable -- and if you pay more for electricity (like $0.30 per kWh), you'd be breaking even at just 1 W/GH!
Anyway, for those thinking the current levels of efficiency were the end of the road for Bitcoin ASICs, there's still plenty of room left for optimizations. The first wave is now over (and probably the second and third as well), and the focus is now on refining designs rather than just getting them out the door. It's going to be interesting to see what sort of pricing we get on the next generation of ASICs, but we're still a few months away it looks like.
Subscribe to:
Posts (Atom)