4421
|
Bitcoin / Hardware / Re: tldr: Why isn't KNC able to compete with *55nm bitfury's W/GH?
|
on: September 13, 2013, 06:49:16 PM
|
One might also ask: Why does BFL's 65nm chip consume ~8x as much power per hash as BitFury's 55nm chip?
Both are supposedly Full Custom designs, I believe.
Full custom has no standard definition. It is a marketing buzzword these days. What one companies calls "full custom" another company may not. A lot depends on the quality of the design and nobody hand places each transistor so custom or not there is some library used, some software is more granular and some is more macro so it is more shades of gray then a black and white distinction. That shouldn't take away from Bitfury their design is very efficient. One thing to consider is that Bitfury uses a rolled design and all other designs to date have been unrolled. Unrolled design has been the conventional wisdom back to the FPGA days, so Bitfury going with a rolled design (which they did on their FPGA as well) may be part of the "magic". Bitfury likely would have seen commercial success in the FPGA space if BFL hadn't resorted to using underhanded tactics. BFL announcing ASICs would be available in a few months back in the summer of 2012 killed further demand for FPGAs. Still Bitfury FPGA design was very efficient compared to competitors so it shouldn't be much surprise that their ASIC implementation would be as well. Whatever the reason, they use die space very efficiently. A different way to look at it is how much die space it takes per GH at nominal clockspeed. Transistor size is directly proportional to process node so I normalized all results to 55nm (native density * (process node/55)^2 ). KNC & Cointerra not included because they haven't released die size details. Die Efficiency (GH/cm3) Process (nm) Perf (GH/s) Die Size (mm2) Raw Normalized (55nm)* ASICMiner 130 0.33 21.70 1.5 8.5 Avalon 110 0.28 16.13 1.7 6.9 BFL 65 4.00 56.25 7.1 9.9 Bitfury 55 3.00 14.44 20.8 20.8 Hashfast 28 400.00 324.00 123.5 32.0
Note: This should't be taken as an endorsement but more an academical look at how much silicon each design requires to accomplish the same thing if they were fabricated at the same process size. Everything else being equal less die space per unit of hashing power is better. It means less silicon cost and compensating for clock frequency and voltage less power usage. * Smaller process node is going to have smaller features. By calculating the change in die size if fabricated at 55nm it provides a better comparison of die efficiency. Now this isn't a perfect comparison because clock speeds will vary based on process node and how "hard" the manufacturer decides to push the chip. To do better though requires knowing the clock speed of each chip and the corresponding performance which I don't have or even know if it is available.
|
|
|
4422
|
Alternate cryptocurrencies / Mining (Altcoins) / Re: Swedish ASIC miner company kncminer.com
|
on: September 13, 2013, 06:08:09 PM
|
P.S. The power supply looks like a Cooler Master V1000. Or any other model from V series, they all look the same. For example V850 Given they recommended to user 850W I would assume they bought V850s for internal use. I would assume they didn't buy smaller PSU for the smaller rigs and instead just plan to use 1 PSU for 1 Jupiter, 2 Saturn, or 4 Mercury.
|
|
|
4423
|
Bitcoin / Hardware / Re: HashFast announces specs for new ASIC: 400GH/s
|
on: September 13, 2013, 05:10:06 PM
|
HashFast,
Some questions:
Can you provide the dimensions of the package (important as I am looking into immersion cooling so critical heat flux is important)? Can you clarify that the nominal wattage of the chip itself is ~250W and the wattage of the system at the wall is ~350W? Both numbers have been used but it isn't exactly clear what they represent. Can you provide the dimensions of the ASIC board? Estimate is fine. Can you also provide an estimate of the height of the tallest board component (excluding waterblock)? Can the controller handle more than 1 hashing board? If so is there an upper limit? 4 boards? 10 boards? 50 boards? Will you consider selling just ASIC boards instead of complete systems?
|
|
|
4424
|
Bitcoin / Bitcoin Discussion / Re: Bitcoin is slowly shit.
|
on: September 13, 2013, 04:43:19 PM
|
Pretty lame if you ask me - especially when bitcoin touts itself as being "free of transaction fees". Sooo, you don't have to pay a fee, but it will take over a day for confirmation - gay.
LTC has minimum mandatory fees (mmf) on low priority txs just like Bitcoin does. The irony of your statement is that the mmf for LTC is about double that of BTC. Minimum mandatory fee for low priority txs (high priority tx have no required fees on either network) LTC: 0.01 LTC (10 mLTC) = 2.5 US cents per KB BTC: 0.0001 BTC (0.1 mBTC) = 1.3 US cents per KB Facts. I know they can be annoying.
|
|
|
4425
|
Bitcoin / Bitcoin Discussion / Re: Bitcoin is slowly shit.
|
on: September 13, 2013, 04:28:21 PM
|
That raises another interesting question: how to make that broadcasting stop in case the sender realizes he has set the transaction fee too low and wants to correct it?
It can be done but it should be considered a hack as a last resort. You will need to use the pywallet tool and delete the transaction from your wallet. You will want to make separate backups BEFORE and AFTER the update. Then you will need to ensure you node stays offline until other nodes forget about it. If your node immediately goes online there is a chance another node will relay your tx back to you and your wallet will happily include it in its history (essentially undoing your erase so that is why I said make a post edit backup). Still it can be done. Eventually the entire network will drop your old tx and without your node to "remind them" by rebroadcasting it will be like the original tx never existed. You can then create a new tx. You may wonder why doesn't the client just allow you to cancel. Due to the way Bitcoin works you have no control over what other nodes do so while you can trivially "undoing" a tx locally you can force other nodes to also perform that undo. This means your node will be out of sync with the rest of the network. It is a common abstraction to think of "the network" as this unified system but really it is more like a sea of independent nodes which speak the same language. So what kind of weird problems could occur if there was a "cancel" button in the client. a) You can cancel a tx locally but it still end being confirmed in a block. b) You can cancel a tx locally but your peers reject your modified tx because they still see the original as valid and thus the replacement is a fraudulent double spend attempt. c) You can cancel a tx locally and create a new one but it happens to use different coins. Both tx end up being included in a block and you double paid. d) You can cancel a tx locally and pay a different person with those coins but the original tx confirms and now to the second person it looks like you performed a double spend. Starting to see while you can technically cancel a tx locally (just delete it and stop broadcasting it) allowing users to do that is likely going to create a huge amount of confusion and chaos. Still this entire issues has less to do with fees and more to do with the person running the casino has no clue what they are doing. Not only did they send low priority tx w/ no fees (which means many nodes won't even relay them due to "spam" prevention) but the tx had NO CHANCE of being included in a block even with a massive fee. The reason is that the inputs for the tx (all inputs are outputs of prior txs) are from txs (parents) which aren't confirmed either and one of those has the same issue (grandparent tx is unconfirmed). The operator is utterly incompetent. Worse the default client prevents you from making these types of mistakes. You can't spend unconfirmed outputs and you can't bypass the min mandatory fee on low priority txs. So the operator is using raw txs. Raw txs plus incompetence is a good way to lose Bitcoins forever. You can create raw tx which result in the permanent loss of Bitcoins. It is a tool for experts only, experts who have done extensive testing. As for "I want to hack my client because I don't want to pay fees" mentality. Really this is cutting off your nose to spite your face. The "min mandatory fee" ONLY applies to low priority txs and (post v0.8.2) is 0.1 mBTC or about $0.01. It exists as an anti-DOS mechanism so even if you bypass it by hacking the client (or using raw txs) there is a good chance not all tx wil be relayed by all nodes. The simpler longer term solution for unconfirmed txs is for something which has been proposed called the " child pays parent" patch. It doesn't need to change the protocol just how miners pick txs. Currently miners will only include txs in a block if the inputs are already confirmed. So you can't confirm a tx (tx A) which won't confirmed by spending it (tx B) because even if tx B has a fee no miner will include it until tx A has been confirmed. With "child pays parent". Miners will look backwards to find the unconfirmed inputs and include them as well. So if miner accepts tx B (which has a fee) and tx B has tx A (unconfirmed) as an input then the miner would include both in the block. Of course that still won't solve general incompetence which is the larger issue not the network.
|
|
|
4426
|
Bitcoin / Hardware / Re: Novec 7000 Project [immersive evaporating cooling]
|
on: September 13, 2013, 04:16:26 PM
|
So its $1k to cool this container. Sinks and fans will be cheaper. What is the business model for the fluid ?
The goal isn't to fill the container and for it to be economical you are going to want to stack the head loads very tightly. Most engineered fluids are ~$50 per Liter. 1L = 1000 cm3. So we are talking $0.05 per cm3. Now imagine you have a KNC board you wish to cool. This isn't a recommendation however it is the type of board that would be a good fit because it has a high energy density. The board is roughly 15cm x 15cm. Now put the board on its side and put a 1 cm spacer between it and the next board. Do that for 10 boards. You end up with a cube roughly 15cmx15cmx15cm. The fluid would fill the space between the boards a rectangle roughly 1cm x 15 cm x 15cm. So there will be 225cm3 of fluid between the boards. It is actually less because the surface mount component will displace some of the fluid. 225 cm * $0.05 = $11.25 per board. Now that is just the cooling fluid you will need condenser and if the condenser is water cooled, pumps, tubing, and either a chiller or dry tower. However this simplistic example should show how while the cost per gallon is high the goal is to use as little as possible. It also allows incredible energy densities. In the example above 10 boards (1,000 GH/s) occupy a space of 3375 cm3 or 3.4 cm3 per GH/s. Now you still need space for the vapor height and the condenser so lets triple that to roughly 10cm3 per GH/s. You could never air cool KNC boards with only 1cm on clearance between boards. KNC stock design is for 4 boards (400 GH/s) in a 4U rackmount space. So roughly 100 GH/s per 1U. A datacenter rackunit is 19" x 24" x 1.75" = 798 cubic inches. To use the same units that is ~13,000 cm3 or 130 cm3. 10 cm3 vs 130 cm3. A 13x improvement in density. While you wouldn't need to rackmount these you could fit about 50 TH/s of KNC gear (90KW) in the same space as a datacenter rack. If Cointerra or HF can deliver 4x the efficiency it would be more like 200 TH/s per datacenter rack.
|
|
|
4427
|
Bitcoin / Bitcoin Discussion / Re: Fragmentation of the Bitcoin network
|
on: September 12, 2013, 11:45:24 PM
|
It will only be less and less if the rate of increase (not the increase) goes up. To get to 1 minute would require the hashrate to rise by a factor of 10 in one difficulty period. It would then quickly reset within a few days unless the difficulty again increased by a factor of 10 continually.
|
|
|
4428
|
Other / CPU/GPU Bitcoin mining hardware / Re: what will replace asic?
|
on: September 12, 2013, 11:12:49 PM
|
Maybe quantum asic, but today's quantum computer from Dwave require huge amount of power and cooling which is not practical for mining usage, and the speed is not that impressive either, about 3600x faster than intel q6600, 39.6Ghs/s
No its speed is 0.0 Gh/s
|
|
|
4429
|
Bitcoin / Mining speculation / Re: If Bitcoins Go Up Will USB Bitcoin Miners Be Profitable?
|
on: September 12, 2013, 09:10:44 PM
|
So what price would a USB maybe be profitable? BTC.1, BTC.05, BTC.01?
At the moment... BTC.01 or less. .085 after 6 months Thanks for disproving your own claim. Even assuming no slowing in difficulty growth, in the next six weeks and waiting a week for delivery, and subtracting 0.05 BTC for power cost the unit would produce 0.08 BTC net revenue. Clearly 0.08 BTC is > "0.01 BTC or less".
|
|
|
4430
|
Bitcoin / Development & Technical Discussion / Re: Checkpoints do protect Bitcoin, don't they?
|
on: September 12, 2013, 08:57:58 PM
|
You could store the RIPEMD-160, WHIRLPOOL, and SHA-3 hashes of checkpointed blocks and for a node to accept a block as valid it needs to not only have the proper difficulty SHA-2 has but much the three stored checkpoint hashes as well.
What approach is used in your Satoshi's client? Only SHA-256? Yes. It stores the blockhash used by the network for validating difficulty (mining and all that). It doesn't have to. It is just a client side check. Hell you could make the checkpoints MD5 if you wanted to.
|
|
|
4431
|
Bitcoin / Bitcoin Discussion / Re: Bitcoin is slowly shit.
|
on: September 12, 2013, 08:55:51 PM
|
Pretty lame if you ask me - especially when bitcoin touts itself as being "free of transaction fees". Sooo, you don't have to pay a fee, but it will take over a day for confirmation - gay.
I'm definitely holding onto my litecoins.
You do understand LTC uses the same model. If you have a low priority tx and don't pay a fee the tx might not be relayed. Even if it is relayed if LTC ever has anywhere near the volume that BTC does miners will prioritize tx based on fees with predictable results. Of course in this case the OP was sent a tx which had as its inputs tx which are unconfirmed because they also are seen as spam by the network. It would be kinda like paying for a candy bar in with a $20 bill and the cashier punches you in the nuts and burns your change. Somehow you would reach the conclusion that fiat sucks and you are sticking with barter.
|
|
|
4432
|
Bitcoin / Development & Technical Discussion / Re: Checkpoints do protect Bitcoin, don't they?
|
on: September 12, 2013, 08:46:57 PM
|
Sure. However if SHA-2 can be trivially preimaged then essentially everything from secure communications, to your Bitcoin address, to SSL to password hash tables are also useless. Since if that happens it is pretty much game over it probably doesn't warrant a backup however there is nothing that requires checkpoints be SHA-2 hashes or even a single hash. You could store the RIPEMD-160, WHIRLPOOL, and SHA-3 hashes of checkpointed blocks and for a node to accept a block as valid it needs to not only have the proper difficulty SHA-2 has but much the three stored checkpoint hashes as well.
Still like I said if SHA-2 is subject to that kind of preimage at will attack well it probably doesn't matter. It would be like asking if an attacker was immortal, could destroy enemies with his mind, and teleport at will would national armies still be effective?
|
|
|
4433
|
Bitcoin / Mining speculation / Re: If Bitcoins Go Up Will USB Bitcoin Miners Be Profitable?
|
on: September 12, 2013, 08:31:44 PM
|
So what price would a USB maybe be profitable? BTC.1, BTC.05, BTC.01?
At the moment... BTC.01 or less. Seeing as a USB BE would generate more than that in the first week I think that is dubious. At 0.1 BTC (or less) is is almost certainly profitable. At 0.15 BTC would likely take a long time (6+ months) and would require cheap power and difficulty growth slowing significantly in 2014. At 0.2 BTC I don't see turning a profit under any conditions.
|
|
|
4435
|
Other / Beginners & Help / Re: What will happen when all the bitcoins are mined?
|
on: September 12, 2013, 03:56:54 PM
|
As I understand, bitcoins will die with last bitcoin produced whenever how long it may take. As to make any transaction bitcoin MUST be produced, just means without producing a bitcoin no transaction can be made. Correct me if I missing something.
You are mistaken (but don't feel bad it is a common mistake). Let me ask you a question. If all the gold in the world was already mined couldn't you still use the existing gold as a currency? Mining will never end (unless Bitcoin is abandoned) however the block subsidy (initially 50 BTC per block) will decline towards zero over time. It is important to think of the "new coins" as a subsidy. They aren't a "free reward", you ARE paying for them right now. You pay for them because every new coin reduces the value of existing coins. The network has real cost and it has to be paid somehow. A subsidy doesn't make it free for users anymore than if a government eliminated taxes and just printed enough extra money to cover its costs. Miners are compensated with both the block subsidy & transaction fees. Satoshi always intended for fees to make up an increasing portion of mining revenue over time.
Remember: 1) Operating the network has a real world cost. 2) As a user YOU pay a portion of that cost regardless of if it is in the form of fees or subsidy 3) Over time the the share of the network paid by fees will rise and the share paid by subsidies will decline.
|
|
|
4436
|
Other / Beginners & Help / Re: What will happen when all the bitcoins are mined?
|
on: September 12, 2013, 03:46:48 PM
|
Then why are you here? You have decided that end state is an inevitability so it is only a matter of time before Bitcoin is dead. So why are you here? Then again if you are going to be here don't make stupid claims that 51% of miners can change the reward especially not in the noob forum. No reasonable person looking at your original claim would "know" your true meaning is that in the future Bitcoin will be centrally controlled. You didn't say the Bitcoin Central control agency will force the block subsidy higher, you said 51% of miners will choose to pay themselves more.
|
|
|
4437
|
Other / Beginners & Help / Re: Crypto Compression Concept Worth Big Money - I Did It!
|
on: September 12, 2013, 03:45:05 PM
|
Oh, duh, I should have seen that one, I would be using characters, which are themselvs ASCII binary, meaning each character would be 8 bits, so I would be adding more data. So can anyone tell me what is going on with computers that can use something other than binary? (of course that's not helpful to this discussion, but I am curious). Is that what quantum computers do, compute with more than binary, but some new base number set? What exists that would be faster than binary, is there anything?
There are no non-binary computers but in theory you could develop one it wouldn't improve storage though. Once you abstract everything away it has a physical manifestation (magnetic charge on a disk, high or low signal in ram circuit, transistors in silicon). Using trinary would be possible but everything would take as much space. For example most flash ram right now stores 1.5 bits per cell reducing the cell count by 1/3 for a given amount of storage space. As for quantum computing, no it also works using binary however it involves superposition. In classical mechanics a bit (or lightswitch if it helps to visualize it) has only two possible states 1=on or 0=off. Either state can exist but only one state at a time. In quantum mechanics a qubit can be simultaneously both on and off. This concept is expanded to larger sets of bits. For example: Consider first a classical computer that operates on a three-bit register. The state of the computer at any time is a probability distribution over the 2^3=8 different three-bit strings 000, 001, 010, 011, 100, 101, 110, 111. If it is a deterministic computer, then it is in exactly one of these states with probability 1. However, if it is a probabilistic computer, then there is a possibility of it being in any one of a number of different states. We can describe this probabilistic state by eight nonnegative numbers A,B,C,D,E,F,G,H (where A = probability computer is in state 000, B = probability computer is in state 001, etc.). There is a restriction that these probabilities sum to 1. http://en.wikipedia.org/wiki/Quantum_computer
|
|
|
4439
|
Other / Beginners & Help / Re: What will happen when all the bitcoins are mined?
|
on: September 12, 2013, 03:36:25 PM
|
If anybody tries to mine a block with too high of a reward, then that block is just ignored by the rest of the network.
Exactly. That wouldn't happen if 51% decided to accept them instead. False. 1%, 51%, 99% it doesn't matter you can't change Bitcoin you can only fork it. When you do there will be two incompatible forks. a) Original Bitcoin with falling subsidy high user support, and broad merchant acceptance b) Inflat-a-coin with massive subsidy, negligible user support, and non-existent merchant acceptance Nobody but miners win from switching to b so nobody will. B will simply die off and the miners foolish enough to switch will mine lots of worthless coins leaving the "loyal" miners who remain on the original chain richer.
|
|
|
4440
|
Other / Beginners & Help / Re: What will happen when all the bitcoins are mined?
|
on: September 12, 2013, 03:33:38 PM
|
If all the BTC Users don't switch to their fork they would simply mine a new coin that nobody uses.Worth = 0. While the miners that staid on the old chain would get a nice chunk in transaction fees due to limited competition.
The users wouldn't have to switch.w reward structure works. I could decide to mine for myself, and pay myself 100 BTC per block instead, but noone else would accept my blocks, so there wouldn't be much point. But if over half of the network decide to accept those blocks, those blocks get built into the blockchain, and the new reward structure works. No that is not correct. Some miners could make the reward 100 BTC and it would be rejected by the nodes of every non-miner who doesn't switch. A 100 BTC block right now is INVALID. It doesn't how many people are mining them. Hashpower can't change the rules. 51% has no relevence. 1% of miners could mine 100 BTC blocks but they would be INVALID on existing nodes as would 99% of miners mining them. The only way a 100 BTC block becomes VALID is if users upgrade. Of course you will never get 100% of users to agree on anything so if some upgrade and some don't you would have a permanent split a fork. In essence two Bitcoins one which has a decreasing subsidy and rising value and one which has a bloated subsidy and crashing value. It doesn't take a rocket scientist to figure out which one will remain popular. The miners which switch to inflat-a-coin will receive lots of coins, the coins will just be worthless. Likewise the miners who remain on the genuine Bitcoin chain will see their difficulty fall and thus their profits rise.
|
|
|
|