Wow.  Over reaction much, it was a typo. 120 block eruptors.  It has been fixed.

Thanks that eliminates one scenario however it is still unclear if they consider the PSU (not supplied by KNC) as one of the components.  I would imagine not since the ATX PSU efficiency can vary and thus the 1.6 J/GH represents max DC load (i.e 1000 GH/s / 1.6 J/GH = 625W).  If true that is fine I will just assume 90% ATX PSU like I have done for other vendors where PSU is not supplied.  That would make it ~1.7 J/GH "at the wall" (625W DC / 0.9 = 694W @ 120Vac).

Either number is better than what I am reporting now but I would like to be correct.

I don't know.  Maybe we need a bot that once every 20 posts on any ASICMiner thread it autogenerates a post containing commonly repeated nonsense.

Well ASICMiner just gave mini rig buyers a punch in the gut.  

Buy from BFL in August 2012. 1x mini-rig 1.5 TH/s @ 3000 BTC
Buy from ASICMiner in August 2013.  120x Eruptor Blades 1.5 TH/s @ 420 BTC
So buy from ASICMiner a year later, no pre-order stress, product will ship in days, spend 86% less, get the same hashing power and receive it sooner.

Things move fast in the mining world so specs only matter if the company can deliver as advertised when advertised.  Those that trusted BFL @ 65nm paid the price, now they want to tell you 28nm will be different.

on edit: Fixed a typo 120 boards not 20.  The typo seems obvious to me.  1,500 GH/s / 1.25 GH/s = 120 (not 20) boards.  420 BTC  / 3.5 BTC = 120 (not 20) boards.

Originally posted in monarch thread to highlight that good specs mean nothing if while you are still waiting someone a year later undercuts it by 86%.  I guess someone felt it was off topic.  It isn't like the monarch is a new product with good specs and who's ROI is heavily dependent on if it can be delivered on spec and on time by the company who still hasn't delivered rigs that people paid 3,000 BTC a year ago that now have a street value of 420 BTC.  Obviously this wasn't relevant at all.

Fixed.

Put it in a paper wallet, locked in a safety deposit box.  If you can buy (or just keep) 100 BTC or buy a miner which will only produce 50 BTC.  It is a pretty heavy "stupid tax" to say "I might spend my BTC so it is better to mine and lose half the BTC guaranteed but maybe the half I don't lose might go up 200% so I can profit". 

If someone has that little willpower well they likely will spend/sell their BTC as they mine it and thus some increased exchange rate won't help.  While mining a rising exchange rate won't help your margins over an extended period of time.  When the exchange rate rises the ROI% rises so more hashing power is deployed. 

The only way the "I hope the exchange rate goes up" miner could benefit (even in the bogus USD metric) is by mining and holding the BTC.  If the miner can mine and hold BTC why can't he just buy and hold BTC?

There is no rational explanation for measuring a device which produces BTC in any unit other than BTC.  If a miner doesn't produce a net revenue (after electrical cost) in BTC which is greater than the purchase price in BTC then the mining was a loss.  Anyone can simply buy BTC instead.

I guess it depends on what you mean by "a while".   In the next 3 months your right power is a non-issue.  That is why ASICMiner/Avalon route was the superior one.  They beat BFL to market despite BFL soaking up massive amounts of cash obtained under false pretenses.


Still the hashrate is relatively low and will continue to grow rapidly in the near term.  5,000 TH/s by December isn't impossible.  At that point using your numbers the Avalon and Bitfury are spending 30% and 3% respectively on power.  For the same amount of hashpower the Bitfury is making 32% more net revenue.

As an extreme example at 6 PH/s (not unreasonable by very early 2014) a user in a high power area (say 30 cent per kWh) would need to spend >100% of gross revenue on power unless the exchange rate rises.  Now you may say nobody spending 30 cents per kWh should be mining and I agree but many people assumed these are ASICs, power won't matter.

So it all depends on how fast the network hashrate will grow.  However I agree for 2013 power isn't really that big of an issue even under the worst case scenario, but 2014 is a whole different story. The nominal numbers don't really matter.  Power was important for GPUs and in time the ASIC powered network will use just as much power as the GPU powered one did (normalized for exchange rate).  There is one critical difference though the difference in efficiency between the best and worst GPU rigs (excluding NVidia) is maybe 3x.  Right now among ASIC devices in the field it is already 8x and if Cointerra meets their estimate it will be more like 12x once they ship.  Normalized for process size Bitfury is the most efficient design, so if a Fury-28* shows linear improvement to efficiency it would be closer to 20x.  That is huge and because of that I would say power efficiency is MORE important for ASICs.

Electrical break even difficulty by device efficiency: https://bitcointalk.org/index.php?topic=281279

* Fury-28 pretty catchy huh? If you like that name guys I wouldn't mind some chips as a gift.

Indeed.  It was more than I was expecting but then again my expectations had been adjusted lower.

If AM can hold the hashrate consistently above 50TH/s this week then the next one should be better.  I don't really know what that multi-week dip in hashrate was but it is ugly and it is going to reflect in dividends.  You can only get paid on hashpower which is operational. 

That would be good to know.  There are adapters which have a switch if KNC isn't going to use the 24 pin connector.  I have soldered plugs with the "power on" circuit permanently closed and I just buy PSU that have a mechanical switch on the back of PSU.  Same thing as paper clip trick but a little more secure and convenient.

My understanding in talking with HashFast is they had some large orders being paid by bankwire that hadn't been paid yet.  They were going to wait till today and then cancel the orders if unpaid.

Hashpower on the farm has improved over the last 24 hours but it was <40TH/s for the prior six days with a couple days averaging closer to 30 TH/s.  That isn't going to help the dividend.  I wouldn't expect anything spectacular.

Each cable looks like this.  There are four of them.  The linked supply has 8 PCIe connectors total.


8 PCIe connectors is actually optimal because while Jupiter only needs 4 connectors and the connector can handle 300W+ there is no easy way to know if a PSU with only 4 connector is designed to deliver that much current on the wire safely.  By the spec a PSU only needs to deliver 75W to  each 6pin connector and 150W to each 8pin connector and be compliant.   Most PSU with only 4 connectors are probably fine but there is no good way other than testing to know for sure.  The nice thing about a PSU like this (or any PSU with 8 connectors on 4 sets of cables) is you know that it can safely deliver 300W (150W + 150W) on each set of wires.

That is their stated goal and as a shareholder I agree it is necessary, but I am also concerned about the ambitiousness of it.  The recent rise in hashing power above 50 TH/s is a good sign.  If they can get the network stable above 50 TH/s for a couple weeks that would be a good sign they have the power and space to handle the upgrade.  Also I think efficiency gain will be lower than 16x.  If it is say 12x then it becomes even more ambitious.    Still the purpose of the thread is to document planned hashpower increases so I have AsicMiner down for 1,000 TH/s in 2013.

If you can show some links or references I will raise the estimate.  BFL isn't the only one who has been delivering hashpower.  I hated in school when teacher said "show your work" (because I could do it faster in my head), but "show your work".

We have always used our own rates that don't necessarily correlate to any particular exchange.  We offer a clients a guaranteed price up front and if they want to sell they transfer the coins to us.

KNC,

Can you clarify what the 1.6 J/GH represents

Is it "at the wall" (assuming 90% efficient ATX PSU)?
i.e.  whole system 625W @ 120Vac for Jupiter (400 GH/s nominal)
625W / 400 GH/s = 1.6 J/GH "at the wall"

Is it the entire system DC load (everything but ATX PSU inefficiency, 90% assumed)
i.e. whole system 625W DC load ~= 711W @ 120 Vac
711W /400GH/s = 1.8 J/GH "at the wall"

Is it just the asic boards (fans additional)?
i.e. 625W asic boards + ~60W for cooling = 685W @ 12Vdc ~= 761W @ 120 Vac
761 / 400 GH/s = 1.9 J/GH

On edit: the word "all components would seem to exclude the third option.

I would like to update the electrical break even point for KNC miners.
https://bitcointalk.org/index.php?topic=281279

Oh and if you are willing to provide an estimate of total pre-orders I wouldn't complain.
https://bitcointalk.org/index.php?topic=278384

Yes.

Very simple solution people have BTC and the are speculating on a rise in the fiat value of that BTC.  They see ASICMiner as a was to "boost" their returns.  If BTC exchange rate triples well you tripled your money BUT if BTC exchange rate triples AND you get 30% dividend you quadrupled your money.

These "investors" are very weak hands.  They are essentially using ASICMiner like a weekly money market funds.   When the exchange rate spikes they want to sell some BTC for USD, except to do that they need to sell ASICMiner shares for BTC first.

Difficulty is not going down.  If someone has expensive electricity their rig is worth more selling it used to someone with cheap electricity then going idle.  Unlike with GPU where rigs could leave the system by people selling GPU on ebay to non miners, ASIC rigs only have one purpose.  So an Avalon isn't going dark until it is no longer break even profitable for even the miners with the cheapest of cheap electricity that is massively higher hashrate.