My understanding (and please double verify) is that a known private key only will not enable you to find another private key, even multiple known private keys won't.  However if the master seed public key AND one private key from the wallet are known it is possible to compute the master seed private key and from that compute all private keys in the wallet. Then again there is no need to ever reveal your master seed public key or a private key so I don't see it as much of an enhanced threat.

Random Wallet
reveal private key - compromise one address
reveal wallet.dat (and passphrase) - compromise entire wallet

Detemrinistic Wallet
reveal private key - compromise one address
reveal wallet file (and passphrase) - compromise entire wallet
reveal master private key - compromise entire wallet
reveal private key AND master seed public key - compromise entire wallet

The first two vulnerabilities are the same.  The third one I just included to be explicit but honestly if an attacker can gain your master seed private key (which resides only in the wallet) it is highly likely your computer is compromised and a random wallet wouldn't provide any more security.

The last scenario is one where a user could (in theory) out themselves.   For example say a user puts master public key seed on a website (so site can generate public keys and a compromise won't result in a loss of a private key).  The user also foolishly gives someone some funds by giving them a single private key.  If an attacker took the known private key and compromised the website to gain the master public key seed then the two could be used together to compromise the entire wallet.  Simple solution don't reveal private keys and if you do generate a new wallet (and thus new master private & public keys) and transfer all funds to the new wallet.

General purpose programmable Quantum Computer is kinda like fusion power.  In the 1980s there were headlines that commercial fusion power was less than 40 years away.  In the 1990s it was less than 40 years away.  In the 2000s it was less than 40 years away.  IIRC just recently a fusion reactor acheived unity (power out = power in) and .... <drumroll> commercial power may be possible within 40 to 50 years.  I know would be willing to bet (even money if anyone is interested) that I won't see a commercial fusion power plant in my lifetime.  This is despite nearly 4 decades of research.  It some way it seems we are further away then scientists thought we were in the 1980s.

SHA-256 (or any other hashing algorithm or symmetric encryption algorithm) is not vulnerable to quantum computing.  The speedup what QC provides against those types of systems is low.   Public key cryptography (ECDSA, RSA, SSL, PGP, etc) is the "vulnerable" area at least in theory.  So the ECDSA used to verify transaction signatures, not SHA-256 used in mining would be the "target" of an ECDSA "attack".  It will take ~4,000 logic* qubits (the number of physical qubits is much more see below if you want to see how far the rabbit hole goes) to implement Shor's algorithm to break a 256 bit ECDSA private key.  Even then it is impossible if the pulic key is unknown.   For the record DWAVE is exactly 0 qubits for the purpose of breaking ECDSA and the largest general purpose quantum computer built to date (IBM) is 7 qubits

Some general purpose quantum computing milestone:
In 2001 IBM researchers were able to factor the number 15 using a quantum computer.  This would be the equivalent of breaking a 4 bit private key.
In 2011 Chinese were able to factor the number 143 using a quantum computer.  This would be the equivalent of breaking an 8 bit private key.

So roughly a doubling of the vulnerable bit strength per decade.  Note this shouldn't be taken seriously but rather is used to illustrate the absolute baby steps being taken.  In both scenarios the amount of time necessary to "solve" these problems with known finite solutions was many orders of magnitude longer than it would take to do it by hand with a pen and paper.  If this doesn't accelerate faster than Moore's law then public key cryptography may never be vulnerable as one can continue to use larger keys.  For Quantum computing to be a threat the capabilities would need to eclipse Moore's law by a significant amount to "catch up" to what is already possible using classical computing.

AFAIK 143 (8 bit number) is the largest number factored used quantum computing.  I think it was Gavin who said (paraphrased) let me know when they can factor a 32 bit number in reasonable time and cost.  Honestly we may be years from even that milestone.

*  As large as 4,000 qubits sounds, quantum computing is very "noisy" and thus to get any reasonable accuracy more than 1 physical qubit is used to represent a single logical qubit to provide a measure of error correction.  This is similar to error code correcting ram using extra bits to correct errors so that number of physical bits on a memory stick is larger than the number of logical bits seen by the computer.   Due to noise the simulation is run over and over and over with the solution of each iteration recorded, the simulation reset and then run again.  Over a large number of simulations the "real solution" will be detectable from the noise.  Very simplified but imagine you used a Quantum Computer and the solutions were represented by letters, the output of a sequence of simulations might look like  A, C, D, A, D, R, F, T, I, L, A, G, Y, J, I, L, K, G.  Since A, D, G occur more frequently it is probable they represent more than just noise, while "Y" for example could simply be the computer recording random noise.  The more qubits used for error correction to more accurate the simulation.  It isn't even really known how much error correction will be needed to get results in any timely manner but one paper I read estimated it at 10 physical qubits for 1 logical qubit.  So we are looking at something on the order of a 40,000 physical qubit machine to break a 256 bit ECDSA private key.  While in theory a 4,000 qubit computer can implement Shor's algorithm unless our material sciences improve to sci-fiction capabilities (building a space elevator would be less of a challenge) it may take an utterly useless number of simulations to identify the true solution from the noise.

Thanks for the datapoints.  It is strange the reported DC/DC output doesn't change.  ~430W out regardless however the input wattage changes significantly.
On edit: fixed misunderstanding of numbers reported. Both output DC numbers are v0.95.  There are no DC output numbers for v0.90.

Need to make some assumptions but lets say the 6 fans use 6W ea (someone can look at the fan sticker and let me know wattage or amps) and the host uses another 5W.  That puts the balance of the system at ~40W.    Lets also assume your PSU is 90% efficient at 220V pretty reasonable for 80 Plus Gold unit over most of its operating range (208V-240V tends to be 1% to 2% more efficient than 120V).

v0.95 =  473W @ 220VAC ~= 425W DC @ 12VDC  (425-40) = 385W Input for VRM (96W per module).

That would put the reported DC output of the module higher (430W ) than the computing DC input (385W).  Something isn't correct.   So either your wall wattage numbers or the output reported by the VRM is incorrect, they both can't be right.  Under ideal conditions (no cooling or host power consumption), 90% DC efficiency, 93% ATX PSU efficiency.  430W output would mean (430/(0.9*0.93) = 513W) >500W at the wall.

Still lets look at the wall efficiency.
v0.90 825W / 495 GH/s = 1.7 J/GH  OUCH.
v0.95 473W / 495 GH/s = 1.0 J/GH

If anyone else has KNC datapoints please provide the following:
KNC Mode.
Average hashrate.
Wattage at the wall.
Firmware version.
# of VRMs present (4 or 8 ).
Power Supply model.
Mains Voltage (120V, 208V, 220, 240V, etc).

Thanks.

Just about all decent brands (and certainly SeaSonic).  The 6 pin = 75W, 8 pin = 150W is a PCIe standard, it doesn't represent the limit of what is possible using the underlying hardware.  The connector itself can handle 288W continually, Molex rated specs for 6 pin Minifit Jr connector is 9A per pin (13A with high current pin).  So 6 or 8 pin PCIe connector is fine for 3 conductors *12 Volts * 9 A ea = 288W  and real world is probably significantly more.  Don't do this at home but I stress tested a connector at 420W for over an hour and it was only warm to the touch.  The wiring itself (3ft, 16 gauge, 12V, 20A per conductor) can handle a lot more so it isn't the bottleneck. 

The only difference between a 6 pin and 8 pin connector is the 8 pin has a pair of ground sense pins to compliant devices (i.e. GPUs) to know which cable is connected on power up.   If you look carefully at many power supply cables you will see there are TWO 8 pin connectors in series on the same cable which plugs into the PSU.  Even under normal usage that means 150W + 150W = 300W on the wires.

What datacenter reboots servers manually?  



I would recommend using a y-cable to connect one switched PDU power drop to both power supplies.



Login to web interface, click power off, power on.  Tada no need to leave your office chair.  If someone was super crazy they would integrate cgminer with the API in popular switched PDUs to auto power cycle when miner appears to be down.  

Note: photos are for NEMA 5-15 outlets (120V, 15A).  Most likely datacenter is going to run on 208V in the US.  If someone was going to install a dedicated branch circuit at home for miners no reason to not go 240V.  Double the power for the same current, higher efficiency, and lots of used PDU on ebay for cheap.   Still the same concept applies at 120V, 208V, or 240V.

Capability is different than sales.   Cointerra like HashFast is using third party to handle assembly and shipping.   How large of a run of chips did Cointerra pay for (or I guess I should say will pay for)?  Probably more than they sold but it is risky to pay for magnitude more chips than you have orders for.  So even if they can ship 25,000 units per day well that still won't be 18 PH delivered in a week unless they have 18 PH of chips (and pcb, and other components) ready to move.

Also Cointerra isn't shipping until January 1st (unless delayed tapeout still isn't done).  By January conservatively the network will be more like 8 to 10 PH/s.   So you are now talking about deploying 50 PH/s or more in a week to cause a max increase.  It could happen but I stand by my prediction that we will never again see a max difficulty change (+300% increase) and the further we get into the AISC "boom" the less likely it becomes.

I had to as well after that last post I doubt he will listen but at least others will get the correct info.

Try to learn something for once.   The efficiency of the chip isn't the entire power requirement of the entire system.

The reported efficiency AT NOMINAL HASHRATE is ~0.65 J/GH.  So at 400 GH/s that is 260W @ ~1VDC (actually voltage depends on chip spec and hasn't been reported but likely is 0.7VDC to 1.0VDC).  An ATX power supply delivers high current on the 12V rail, however no AISC runs at 12V and thus the 12VDC has the be converted to ~1VDC and conversion means energy "lost" as heat (otherwise the VRMs would be cold to the touch).  A quality high current 12 VDC to sub 1 VDC regulator is going to be ~90% efficient so that is 288W @ 12VDC per board. 288W @ 12VDC in = 260W @ ~1VDC out + 28W as heat.

Sill 288W isn't the whole story either.  The watercooling loop requires power as well.  How much?  Not sure but lets guestimate.  Each module includes 2 fans likely high RPM so lets go with 12W ea and a pump say another 6W.  So 30W per cooling loop plus 288W per board is 318W total.   Lastly there are a pair of exhaust fans.  Lets guesstimate 12W ea.  318*3 + 12*2 = 978.  So 978 not 780 watts per rig.  

At nominal hashrate (400 GH/s) & 0.65 J/GH.
Per chip -  260W @ 1VDC
Per board - 288W @ 12VDC
Per board including cooling - 318W @ 12VDC
Entire System including exhaust fans - 978W @ 12VDC

However that is only at nominal 400 GH/s hashrate.  Hashfast has indicated they believe the chips can be pushed harder.  How hard?  Well I certainly don't know and HashFast won't even know for sure until they get the final silicon but lets say they can run at 450 GH/s at nominal voltage and with a 10% overvolt can be pushed to 500 GH/s.  In Silicon devices, power consumption increases by the square of the voltage increase.  So 10% higher voltage = 21% higher power or 0.78 J/GH.
 
500 GH/s @ 0.78 J/GH  (AS AN EXAMPLE ONLY)
Per chip -  390W @ ~1VDC
Per board - 433W @ 12VDC
Per board including cooling - 463W @ 12VDC
Entire System including exhaust fans - 1,413W @ 12VDC

So designing around a 780W power supply would limit the potential of the device.   Maybe 500 GH/s per chip isn't possible but maybe 440 GH/s or 465 GH/s is.  HashFast stated there were no power supplies in the range they were looking at (1300W to 1400W DC) from vendors they were interested in.   Also stated was that even if one was available it was more expensive than two smaller units.  

SeaSonic's "X series" for example maxes out at 1250W and a 1250W unit has higher price than a pair of 750W units (1500W total).

Just use a switchable PDU.  They work great for home "farms" as well.


Agreed.   

Well HashFast didn't design the system for redundant power supplies (it would require much higher output power supplies = higher cost) but pulling double load from one power supply won't necessary "bad things".  It would cost more but if each power supply had sufficient capacity to handle the full load then the system could operate just fine on one.  When both PSU are operating the load on each one will be half.  If one shuts down the other will handle the full load.  You gain redundancy, and higher efficiency at the expense of more power supply cost.  If one PSU fails the other will run at 100% load until the first PSU is replaced.



  

The network allows up to a 300% increase (and a 75% decline) per adjustment.  A max difficulty jump has happened a few times at the start of the "GPU era".  I doubt we will ever see a 400% increase again because the network is so large now.   With the network at 3 PH/s it would need to increase by 9 PH/s within one difficulty adjustment.  Due to timing it probably would need to be be more like 12 to 18 PH being deployed within a week.  That would be like shipping 6,000 BabyJets or KNC Jupiter's every day.

"against the law at state level" is also against the law at the federal level.  You aren't compliant with FinCEN if you are in any violation of state law/regs.  Being registered doesn't make you compliant it simply makes it easier for the feds to find you if/when you aren't compliant.  Being compliant makes you compliant.

ECDSA is not the same as Dual_EC_DRBG.   The vulnerability is with Dual_EC_DRBG not the entire ECC concept.  Actually the speed at which the crypto community sounded the alarm on Dual_EC_DRBG should be seen as a positive sign.  It was/is an obscure algorithm with no real widespread usage and the flaw was found and published internationally in the span of a few months.   

crumbs there is a finite amount of available space to make a rackmount unit.  You design a rackmount unit with better airflow, buy modules from HF in bulk and then resell the package.  I think you will find it is harder than it looks.   Putting radiator in the back would be an "easy" solution except you only have 17" by 1.75 x U height inches to work with.  If the power supply is mounted on the back and the radiator is mounted on the back the radiator will be tiny, too little surface area to effectively cool 750W+.    To keep Delta T less than 10C over ambient you are going to need 1 to 2 cm2 of radiator surface area per watt (i.e 420cm x 120cm on 750W heat load) even with pretty high extreme airflow (3000 RPM pusher & puller fans).  There is only so much surface area on the back or front panel of a rackmount unit.

Sure if you don't want to compromise then build a massively expensive 6U chassis with straight flow power supplies and the entire rest of the back panel devoted to a radiator.  Of course when you do so you would price yourself out of the market and people will just buy the more economical solution from Hashfast or Cointerra.

Unsubstantiated claims. 

HF reported signing a deal with SeaSonic.  The deal said nothing about a custom run.
Some people have speculated the PSU fan operates in reverse.  HF hasn't stated that.

Extremely likely the PSU are the same exact SeaSonic models you can buy from newegg.

ATX PSU are designed to intake from heated case air.  It isn't being too lazy.  A rackmount case is only 17.5" wide.   A power supply is 3.4" wide.  You would need to use a more expensive server (designed to have front to back airflow) power supply and then you would still lose 20% of your front surface area.   Removing 750W+ by watercooling is no small task and that means a relatively large radiator surface area.

Sure if you wanted to make the case larger 6U+ or only put 2 Sierras per case but those would be inferior choices IMHO.

Power supplies are designed to work at high ambient temps.   Both servers and PC intake cooling air from inside the case.   Some high end PC allow flipping the PSU to draw outside air but that is the exception not the rule.   SeaSonic puts a 5 yr warranty on their power supplies and they know 90%+ of the time it is going to draw in heated case air.  They are designed to handle that.  Actually it has become an easier engineering challenge as PSU become more efficient it means they have less of a heat load.

The PSU (like evey PSU made) exhausts out the "back".  If installed as shown in HF model it would intake from the outside and exhaust out the back.  If you flip the PSU 180 it would instake from the inside and still exhaust out the back.

Every single rackmount server pulls in warm case air to cool the PSU.  PSU these days are highly efficient they don't need ultra cold air or high RPM fans.   A low airflow of hot (90F+) air is well within design spec.  It is a non-issue.  At 600W load and 90% efficiency you are talking 54W of heat.  It is more expensive (due to the amount of metal in larger heat sink) but they make passive power supplies because 54W in a space the size of a power supply is nothing it is less than a lightbulb in a space 10x larger.

The power supply is a non issue.

Outside of datacenter.  Use it as shown in the photo.  Draw in cooler outside air, exhaust out the back.

Inside datacenter (w/ restricted side access).  Flip PSU 180 degrees.  Draw in warmer case air, exhaust out the back.

As far as PSU can't handle the heat load.  It will handle it just fine.  MOST PSU draw in warm case air.  Take a look at any rackmount server PSU.  They ALL draw in heated case air.  In a datacenter the power is likely 208V.  At 208V/240V PSU tend to be 1% to 2% higher efficiency.   At 90% efficiency you are looking at 50W heatload.   It is like the ultra difficult task of cooling a lightbulb in 90F air.   You know like what happens every single day outside in the summer. 

SeaSonics are built like a rock.  Open one up or look at reviews which show internals.  Massive heat sinks, giant transformers, oversized Japanese caps.  I ran a farm of GPU in a 90 deg garage for better part of two years using SeaSonics.  Not a single one died, not a single one underperformed.  We are talking 1200W units, 90F ambient,  dusty conditions, 95%+ load.  They just work.

Sounds good.  Will PM you.