Topic: [$0.35/GH] Bankrupted HashFast Hardware Sale (Read 5376 times)

In the normal financial circles this would work as follows:

1) group buy collects the funds in a bank A's account
2) bank A issues a letter of credit
3) bitcrane takes the letter of credit an obtains a loan against it from bank B
4) using loaned funds bitcrane manufactures the miners and ships the miners
5) upon proof of shipment bank A releases the credited funds to bitcrane
6) bitcrane repays the loan to bank B

In other words:

a) bank A is supervising the availability of the funds
b) bank B is supervising the timeliness of manufacturing and shipment

But this is Bitcoin, there's nothing normal about it. In particular many Bitcoiners are actively against banks and conventional finance.

The onus and majority of risk should be on the manufacturer/supplier building the machines, the xx% deposit idea shows the manufacturer the customers are serious about purchasing. Using an existing known manufacturer, as is the case here, should add credence to the deal, however the weight of responsibility needs to move back to a normal supply chain system where companies are shipping from stock.

Well, if someone wanted a quick % out of a half million dollar investment, maybe this could happen.

I think that one way this would work would be if buyers could pay like a deposit, 10% to secure their miners and then when the miners are built and ready to ship they could pay the remainder.

Then Bitcrane should build them and sell them from stock. Anything less is unacceptable.

oh yes please gimme gimme wait no.

The thread about their already existing Hashfast powered rigs is here: https://bitcointalksearch.org/topic/bitcrane-announces-black-friday-sale-718383

I don't keep track anymore of the things that I "don't like" at HashFast, to use an euphemism. I don't think it could suck more.

Anyway, in order to try to raise trust towards the manufacturer Simon was authorised to share their name; the company is Bitcrane, and they do their manufacturing in Shenzhen.

I am very suspicious of this and would not want to buy from them. I have and would buy again from pepperming but not from Hashfast.
The pepperming boards are great and are indeed very flexible when it comes to changing voltages and clock speed to get the power to hash rate you want.

I would rather buy S3’s than boards from HashFast.

Well it might comfort some potential customers that a portion of the money they spend will indeed be used to maintain the comfortable meager lifestyles of the two people mostly responsible for fucking everything up.

It wouldn't comfort me but hell I'm kinda funny like that 

yes, i have a prejudice against:
a) a bankrupted company
b) a company with a terrible history of non-delivery
c) pre-orders

when you combine all of the above to create the new hashfast - its more unappealing than ever.

the price and efficiency is okay - but there is not a chance in hell that i would trust them with a pre-order, especially of such mainitude. This is not a company that anybody should be trusting half a million dollars to on a pre-order.

now, if someone wants to fund this, and sell boards that are in stock today, I would be interested. But right now hell is still hot and I am not giving any money to hashfast.


I wasnt fooled the first time as I watched others struggle and complain - and I dont plan to be fooled this time until theres proof that boards work and are shipping

Sucks about this whole HF ordeal, their ASIC is truely phenomenal. I have a few myself, extremely flexible clocks & voltages is what I enjoy most about them.

You can hear it in the hearings and have it confirmed this Friday (in the status update hearing). Simon and Edward are still there, but they signed most of their powers away to the CRO, Victor, that was agreed with the committee. Edward was moved to sales but kept his full salary, at least the last time I heard something about him. So yes, they are still being paid 6 figures, and that was the case even before that they raised their own salary at the beginning of this year (twice) while they weren't shipping machines due to a lack of cashflow.

Ya I saw that but it is also being heavily touted that old management is no longer involved.  I am hoping that Eddie and Simon no longer hold C level positions within the corporation.  I assume C-level execs are considered management.  Also if they are still there couldn't they have different titles so they didn't have to pay them 6 figure salaries?

Not exactly a definitive, proven statement but https://bitcointalksearch.org/topic/m.9287659 I'd be interested in knowing the formal answer to this too.

Are Eddie and Simon still drawing salaries?

Yup. They either need to stop paying the executive officers and redirect those funds to get the equipment in-hand which is the only commercially sensible decision if they want bolster trust and gain a shred of legitimacy, else market their assets directly to someone who isn't a managerially incompetent scam merchant (that counts blackarrow out too), maybe Marto74 might at least stand a modicum of a chance of being able to turn these into working miners.

I am pretty sure his answer is based on the old phrase: "Fool me once, shame on you; fool me twice, shame on me."

As much as I dislike FailFast this is probably the only way customers can get anything back. I would purchase some gear if it was in hand for sure. Not a chance of me pre ordering from FailFast or any other ASIC company though...risk is not worth the reward. I hope the people that got screwed can somehow work something out. FailFast were just incompetent mixed in with some shady management. Good luck on this though!

Because of your prejudice for a bankrupted company with a new management and dozens of lawyers' eyes double checking everyone of their moves? Or because you already speculate to know the specs of this other product to be non-competitive?

Because neither one of them make any sense to me.

you are an idiot. he just laid out in very clear terms why his post about FPGAs (which you dragged up from 2 years ago to prove an incorrect point) was correct, and that FPGAs at the time of his post were not actually a good idea. I can back that up, because the GPU values stayed, and FPGAs all went to worthlessness the second ASICs arrived.

point is that hashfast has a terrible reputation, and a poor price. Bitmain ships on time. You are shilling like crazy for godknowswhy

NOOOOPE

I think it to be way less than 30%, and CEO, CTO, CRO are still being paid a total of around $50k a month in salaries, there will be priority claims, lawyer expenses, etc, before coming down to standard claims. So basically I won't be seeing even a billionth of that amount.

So it's not about giving money to HF creditors, since that it's not gonna happen.

In the best case scenario it is about building trust so that HF can take another round of preorder in the future when their rev4/rev5 will be ready. We have paid them to work on it for months, but naturally they won't give us status updates.

I am sure if someone was dumb enough to order FailFast would screw it up. How anyone could even consider them blows my mind. They should just sell in hand chips to people like Mr. Teal / TechnoBit and let them handle it.

I get what you are saying with the fans and watercooling stuff. They do have chips and want to use us to make 2000 boards with there chips. So they gotta be making something in there. Nobody is gonna just give chips away. Yeah the price may be slashed, just to get some value, but I doubt HF would get all these boards built if there is not something in there for HF.

I know you have a rough estimate of what these boards cost to get made. 

They really aren't making any margins, in fact they're almost certainly losing money. It's just about mitigating loss. Right now they have a bunch of stuff with somewhat unknown value (the chips themselves, IP) and some stuff with a known value that's difficult to recover (coolers, fans, cases, long leadtime PSU parts, etc). If (example only, I don't know the numbers so don't take this as a real value) you've paid $200 per kit of stuff and can spend another $150 in order to sell it for $260, you've essentially turned $200 in parts into $110. Yeah you lose money, but it's better than liquidating them for pennies or letting them sit on the shelf for another 6 months.

It's the only way the customers are ever going to see any money from HashFail. Even if it's pennies for the dollars that they've spent it's better than nothing and it seems that this is the legal bankruptcy way after eating 15M$ and producing nothing. (oh yeah lots of failed PCBs and lots of useless and inefficient chips)

Idiot icetard is looking very much like retard Inaba at this point. BFL was claiming that they are a good company just because they have 45k customers, but just like in HF's case everyone is trying to stay away from both companies no matter how much the cock-suckers sock-puppets praise the companies.

How much do you think the actual markup is? $260 is pretty cheap. 2000 x 260 = $520,000.00  HF has got to be making at least 30% margins. $156,000 is probably not enough to cover their customers.

Not. This is for trying to repay HF's customers since HF was unable to do it and all they could do it to make some millions vanish.

Is this basically like giving HashFast another chance? Why would we do that?

Numbers scare you, don't they. Let me make this simple for you.

Buy Bitmain: Get more bitcoins.
Buy bitcoins: Get more bitcoins.
Buy HF: Get less bitcoins.

More bitcoins vs less bitcoins. The choice is yours.

You do a fantastic job of using FUD to make 'perfect' and 'best' the mortal enemy of 'good enough.'

Any mining back then was well worth it.  You can snivel about how your baby was the cutest, but the point is that we may still be in that situation, so you could well be wrong about HF being a "horrible" choice.

You really wanna go there? Ok. Watch and learn.

1x ztex: $1355 for 850 MH/s in Apr 2012. $1.6/MH
2x Icarus: $1138 for 760 MH/s in Feb 2012. $1.5/MH

Icarus boards were cheaper and started mining months in advance of ztex. Ztex was overpriced and late. Hmm, just like HF now. No wonder you like ztex.

PS: A ztex board would have mined about 100 btc over its life. Had you purchased bitcoins instead you would have had over 200 btc. ztex again was the loser.

PPS: GPUs were getting double the performance for the same price with plenty of resell value afterwards. That's who got all the bitcoins back then.

they  might want to reach out to megabigpower or some other company that wants all of it.  If you could put several of these in a case - then surely there should be same way to cut the price of the BOM down so that it makes sense.

Yes, that was great advice when you told ztex not to mine hundreds of BTC with an FPGA, because of your overweening risk aversion.


You are clearly the master of FUD and we should all listen to your wisdom!   

Since you have no actual arguments, you must be conceding that I was right all along. Thanks.

this needs to be less then bitmain's s-3 price on ebay.

which is around 240 -265
 More then 6 or 7 sellers will ships an s-3 to you in about 2 days. And you get ebays return policy if they do not run. and you can pay with a cc.

this deal would be okay if it was 200 with a 100 piece minimum.   but 260 and 2000 pieces it is a no go.

Looking farther down the road,take this into account before buying any mining gear:

http://www.bloomberg.com/news/2014-10-21/bitcoin-miner-ditches-clients-to-chase-2-billion-coding-prize.html

KnC is going to do ANYTHING it can to OWN ALL the bitcoins.............so IMO,BTC mining in any location other than .02 cents kwh colo will be made obsolete very,very soon 

You think they will hold their BTC to keep the price up,not me,they'll dump daily all they make,further suppressing the price of BTC..............

LOL at comparing Bitmain to FailFast...

See?  There's that Manichean hero/villain classification system I just warned you about.

dont put people on ignore, you just miss all the fun.

consider also this when buying old stuff: bitfury is about to make new 28 nm chip.

And to think I almost put you on ignore. You are hilarious! There's nothing "new" about Bitmain. They've done multiple generations of chips and shipped so many units you can't comprehend. No one does logistics better than they do. The few times they did drop the ball they fessed up and offered compensation.

Keep in mind these

1) Doesn't include PSU, shipping, screws or standoffs

2) Comes disassembled

3) No warranty or support because the company doesn't exist any more

4) That price assumes overclocking that not everyone will be able to achieve. Standard performance of 700GHps means over $0.41/GHps

5) Will take *at least* a week before it is shipped, and that is after GB is finalized, I'm guessing these won't be in the customers hands until at least 3 weeks to a month from now


I'd pay no more than $0.2/GHps under these conditions.

Unless Bitmain has some problem with components, logistics, customs, certifications, firmware, drivers, etc.  (IOW all the things that cause HF and the rest to delay).

I understand that Bitmain currently still enjoys a shiny new reputation, but you need to realize that every new ASIC company was seen as the savior until they inevitably dropped the ball on something.

Your Manichean hero/villain classification system is silly, and just sets you up for repeated disappointments. 

No wonder you gave ztex such awful advice.  You discouraged him from mining hundreds of BTC, because you believed your own FUD about "ZOMG REWARD HALVING ZOMG POWER COSTS ZOMG 2 YEARS TO BREAK EVEN."

It's a good thing that not all of use are so short sighted and obsessed with risk aversion.   

Because you lower your profitability. But hey, if you've got so much money that you don't care about profitability, they go for it! You're the perfect HF customer.

Why not order from both HF and Bitmain?

Have fun with your Bitmain pre-order.  I'm sure you'll get all worked up and emotionally overwrought when they inevitably make a mistake.

Not everyone is a Debbie Downer like you.  Many of us want to place bets on BTC testing and exceeding its previous ATH.

Remember when you were discouraging people from mining with FPGAs?   


LOL, thanks for that great advice Negative Nancy!

icebreaker can't understand how come the fastest chip in the world isn't selling. Everyone must be stupid since they are not buying the worlds fastest chip...

Bitmain is shipping sooner, has better power efficiency, is lower cost, comes fully assembled. HF loses on all accounts. There's no intelligent reason to chose HF over Bitmain. That makes it a horrible choice.

Maybe mounting a cooling head to a bolster plate is hard for you, but many of us find it easy and have lots of practice from overclocking our video cards.

Just because the efficiency isn't the best doesn't make it "low."  1W/GH is better than many other ASICs on the market, and you can always underclock to make it better.

$250 for a 750GH Yoli kit is a great deal.  The price isn't "horrible" just because you can name one supposedly superior Bitmain product which may be shipping soon.

cedivad is doing all of us HF creditors a favor by organizing this group buy, and you fling poo at him like some kind of deranged chimp.   

Then it's DOA. There's no place in the market for it.

By the price being right, I mean that it's the lowest possible that can be offered without having the Chinese manufacturer loosing money in this too.
I have no idea of what the right price for hardware is nowadays. I'm not even sure there to be a price. Is $0.01/GH a price? Because that's the price I might be buying at, with tomorrow delivery and 60-days payment.

Unfortunately, it's me. But I'm just a messenger. Again, I don't get a cent off this, I'm only trying to propose what was proposed to me, in the most transparent way possible, since that everyone else decided that proposing this thing to the community was only time lost, so they gave up even before trying.

Sorry cedivad, but the price is horrible for a low efficiency, pre-order, diy kit. Are you sure your BCT account wasn't hacked?

Rather than respond in public, schwab is pestering me with PMs.

LOL, what an idiot.  Makes me miss the newbie jail.

Guys, I completely understand the hate towards HF. My claim is public. Go look that up if you want to know how many $, or rather how many BTC of hate I have towards them. Just run a whois for cedivad.com for my full name to be used in that search. That doesn't mean that they can't do something right, or at least try to do something right, under the guidance of someone different than their old CEO/CFO, the ones that costed HF's creditors more than $15M.

Is the price right? Yes. Does the J/GH ratio still sucks? Absolutely. Does trusting them with a 20-days preorder window sucks as well? Yes. Will they run away with the preorder money? No, since that it would be easier then to lock them away, and I'm quite sure that they don't like the idea, but that doesn't account for the Chinese supplier, of which I don't know the reputation. Is the price the best you are ever gonna get for this specific hardware? Absolutely.

So, while I agree with the sceptics of this thread, while I completely agree with them, please try not to throw your shit at people that are trying to do their best, possibly with no interest, like me (In my specific case, my net gain from a successful preorder is $0.00 and in the best case scenario I would earn something like a tenth of a cent on the dollar on my clam). Feel free to hate IceBreaker and his following shills, feel free to hate the guy who wrote the emails in the first post for his past lies, but please recognise that you are not dealing with them, not anymore.

That doesn't mean "buy", since that again, I agree that some of the alternatives could be more appealing, and that Bitcoin mining is unprofitable in general, that just means to please, not to accuse people who did nothing - in this case, the current bankrupcy management, of having done anything wrong, since that's not the case, and that with a tiny research you could agree with me on that, instead of blindly following and agreeing with the dominant line of thought.

If you had bothered reading the OP before blathering FUD, you would already know the answers to those questions.   

I'm gonna quote all the giant red flags, so far, in the thread.
Bolds are mine

First up, Hashfast itself. Oi. I mean, are you buying chips? Do you have a board design ready? Is it two years ago? Hashfast could be for you, I guess.

Nothing like DIY. Little elbow grease never hurt. I have some competency.

Whoa there, I thought you said it was a kit. I mean, hell at 750ghs that isn't so horrible though. Decent electric...

Wait, what? Easy/Careful. Pick one. Doesn't matter without proper community support this kit will be neither. I'd be hopeful to get 600gh/s.

Well there is the truth. Get to the good stuff. Ok, but we get the bolster and the foil and it's a kit! J/GH seems a little high ya know.

Daddyfatsax-That's what it sorta seems like. Everyone doing a big favor for HF.

Cedivad-The truth! The truth shall set you free from this thread! I'm guessing the, uhh, higher speeds, are the easy 700 or careful 750. It hurts to want to laugh.

I've seen anytime this guy shows up, no go. Not very good. He's weird, a shill. All about the scam and the troll. Just a fucking pain really. Talks out of ass.

Cedivad - Yaaa..... bankruptcy sale, wanting preorders, saying no. Step 1. Take money. Step 2. Run.

You even fucking know.

And a couple decent machines came out, for limited use at high dollars. Open Source designs. Release some shit. Sale reference designs. Oh man, just what a cluster.

I would be interested in 50-100 units (@500 GHS) if someone can figure out a way to get the chips, controller, power, cooling, etc. into a rackmount case for the price at the price and performance point of the C1 or better.  Maybe only bitmaintech has the scale, but just throwing it out there in case anyone (Mr. Teal?) has the ability to source this and maybe we could make an offer for the chips on this basis.  PM me if there is a way and a will.

I'm sure Mr. Teal would mount them on his boards for a small fee.

LOL More chinese garbage trying to be sold as useful, BLAH! Get LOST with this pre-order our crap, CRAP!

And with a board they aren't competitive. Lose lose.

The chips have been on sale for months. However they are quite useless without a board of some kind.

Just sell bare chip guys, this is the best bet for you. Nobody wants the boards, just the chips!

so 260 usd  for a kit  so if I run it at 500gh and .8 watts it is the same as an s-3

I have to pay in btc and wait for this kit.

I go to ebay and for  245 I get this

http://www.ebay.com/itm/Bitmain-Antminer-S3-Bitcoin-Miner-/271642550778?pt=US_Virtual_Currency&hash=item3f3f2861fa


Or 259 I get this

http://www.ebay.com/itm/BITMAIN-AntMiner-S3-Bitcoin-Miner-ASIC-IN-HAND-ships-within-24-hrs-/291272899697?pt=US_Virtual_Currency&hash=item43d137bc71

So I just don't see why anyone would want this gear at this price.

I personally have the BOM for the rev3 board, and I would like you to see it. I'm trying to be authorised to share it. Long story short, there is no way the price will halve, and I personally can't understand how they can get this price in the first place. Even by considering the cooling system free, the rest of the components should add up to more than that.

I think that basically HF could make more money off this by selling the cooling systems at $80, that should be below market price. (too bad that the kits are incomplete, and thus can't be sold as retail).

The C1 ships in a few days. This HF unassembled pile of parts might ship in a month. That makes the HF offering way more expensive.

It's actually the opposite. Mounting the cooler with the Yoli is a bit of a PITA, but it's still simpler than the C1 as it's a closed loop cooler. The C1 you need to source the pump, radiator and coolant and then install and fill your cooling loop. Syscool sells a (super cheap) kit that takes care of a lot of the parts sourcing, but it's still more work to get up and running.

It's not worthwhile anyway to try and run as hard as you can. Running at 975-1000MHz (~750GH/s) is a bit hit or miss and can either work reasonably well or require a big bump in voltage to get to that point. Depending on where you are in the chip lottery you could be running from 950W at the wall to well over 1000W to hit that 750GH/s.

Unless you have extremely cheap power, it's better to scale it back to the 600GH/s or so range, where you can keep the power draw at the wall around the 1J/GH mark.

I would be interested in a quantity order even given the history and bankruptcy, but the price point would need to be a lot lower to be competitive.

The Bitmain C1 is an almost perfectly comparable product - .80 J/Ghs at the wall, 1 Terahash, liquid cooling included... but unlike the Hashfast, it is a fully enclosed ready to run system and is priced at 1 BTC / Terahash.  The GB here would be for a product that needs work, and at .80 J/Ghs the performance is 500 Ghs... so to compare apples to apples, you need 2x of the Hashfast to achieve the same results (at .80 J/Ghs) as the Bitmain product, plus I would have to spring for an enclosure, controller/rpi and spend time assembling things...

That makes the Hashfast bankruptcy sale 2x++ more costly than the Bitmain product! on a comparable (J/Ghs) finished product basis.  Plus, I get top notch customer service and after-sale support from Bitmaintech that would not be available form a bankrupt company.

As a liquidation specialist, I would think that Hashfast's trustee could do a lot better than this.  Taking out the cost and trouble of assembly, case, standoffs, screw, etc. and the lack of meaningful warranty (assume they would do warranty on DOA basis, but in bankruptcy sales even that is sometimes not done) .. I would imagine the price should be half of what it is now. 

Suggest that you look at the market again, and come back with detailed pricing and terms that make sense.  Larger minimums may be needed, but at the very least you should have an offering that beats the C1 pricing by a good margin.

Great point MrTeal.  And most consumer 1200+W PSU are not up to the task of feeding hungry maxed-out Yoli boards for very long until they fail (sometimes taking a board with them)!  You could get two 1200W Seasonics, but will pay a premium for their server-grade reliability.

They will, though you'll probably be limited to two boards per chassis if you want to run the Yoli boards full out, unless you want to invest in a couple 1200+W ATX supplies to power three.

i think he is saying that lol...let me flush some money down the toilet instead - easier way to get rid of it.

The empty cases are bulky, but don't weigh very much.  They are custom cut to accommodate three radiator/fans and two PSU, although I'm not sure if the board mounts work with the new Yoli boards.

Which bitcoin colo won't host S3s?

You are right, and they should be like $50 each. They are in the schedules. However that adds to the complexity, and I guess that shipping it won't be cheap.

I think HF still has a bunch of Sierra rackmount cases.  Even if they don't, bare boards are much easier to put in rackmount cases than the nonstandard Bitmain chassis.

Haha pre orders again? I can not imagine anyone is stupid enough to do this...

Nice idea. I understand HF needs to pay for the hardware, but personally I feel if they want to "make up" with the community they can fork over their own money instead of ours.

Thanks for answering my questions.

Those bare boards aren't rackmountable either. You need to setup them like that on your own if you wanted to rackmount them.

Now please go away.

Please note: the image IceBreaker posted has J/GH measurements taken at the chip level. You need to multiply them by 1.1*1.1, but those numbers vary on the efficiency of external and internal transforms (and could be lower).

I was proposing to keep a 2-of-2 multisig address. I would own the first key and HF the second one. For now HF decided that they are not interested into this escrow account, and that if this thread had enough attention, it would have been something that could be done.

However, this escrow account would be only until for group buy purposes, since that the founds would have to be released to HF that would then pay the Chinese supplier completely upfront.

I think that an alternative, like the funds are hold in the escrow account, while HF pays the Chinese supplier out of their pocket, and I release my key only that the majority of the people successfully receive their hardware, could be worked out, if HF had enough cash, that is something I think not to be the case.

Liar. Bitmain products work just fine in a bitcoin focused colo. The Bitmain S3 is far superior to this risky pre-order crap.

Even AsicMiner is taking pre-orders these days.  No risk, no reward.  Maybe you should stick to buying used machines for $1/GH off eBay instead.



The Bitmain isn't rackmount, so no good for colo use.

The Yoli will do 700GH/s without breaking a sweat.  Wattage varies greatly depending on how fast you run it. 

Well then who would I be sending BTC to? If I decide to buy any hardware. Some form of escrow would be ok, but I doubt that since you kinda wanna do group buy.

The price is good, but this is one of those "too good to be true" kinda deals for me.

No, I don't want anything. I was proposed this deal and I'm doing my best to propose it to the community in the most transparent way possible. HashFast is now under the control of a CRO, so you are not buying from the scammers I talk about on hashfast.org. The price per GH should be one of the best of the market, so do your own calculations. I would probably order if I had cheap energy, just to finally have some hardware from this company, 1 year after the date they promised to ship my Baby Jets.

The board is "up to" 750GH because you need some special (and cheap) components to push it at that level, that are included in the price. Without "bolster plate, thermal pad and indium foil square" you would only get around 450GH per board, that is where $0.60/GH comes from. It's all explained in the emails.

I will edit the title to reflect that.

The J/GH ratio is 0.8J/GH at lower speeds (around 450GH). It goes around 1.1J/GH at higher speeds. (I will now publish the email with that information if it's not in that list already).

I hate "Up To" ratings. What's the guaranteed hashrate? What's the wattage at the wall on these? It's looking to me like a Bitmain S3 would be a far better buy.

So you want us to pre-order stuff from Hash-fast? Am I taking crazy pills?

FAQ Reserved post.

HashFast is one of the few companies that was promising 28nm hardware in 2013. As some of you might now, they are in bankruptcy under chapter 11 and have an external CRO (Chief Restructuring Officer) appointed, meaning that the management is not the same as the one that bankrupted the company in the first time.

I was contacted a few days ago by Simon Barber, telling me that they were finally able to secure a very good pricing for manufacturing on their old but proven rev3 “EVO” board design. It’s a single board single chip liquid-cooled miner that typically can be pushed up to 750GH/s.

Hashfast is offering a kit consisting of:

• A tested good rev3 ‘EVO’ board
• A CoolIT ECO-III liquid cooling system, with Sanyo Denki high speed fan.
• A bolster plate, thermal pad and indium foil square

Pricing:

The price per kit will be $260. Shipping, handling, any taxes or import duties and power supply are not included in the price. To complete the system a PC running cgminer and USB cables will also be needed, as well as common hardware like screws and hex standoffs. The kit will have to be assembled by the customer.

Minimum order quantity:

The above pricing can only be reached if HashFast places an order for 2000 boards with their manufacturer. The goal of this thread is to collect interest on this product, in order for HashFast to be able to place an order.

Boards will ship in boxes of 20 from China (FOB Shenzhen). Coolers, fans, bolster plates, cooling pads and indium foil ship from the USA. The customer will be responsible for purchasing screws, standoffs and PSU needed for final assembly of the mining rigs, and doing the final assembly of the board, cooler, fan and PSU.

Orders must be placed in multiples of 20 kits.

Board performance:

With the provided liquid cooling system, board performance will vary depending on how carefully you place and tighten it to the board, and what mounting hardware is used. Please refer to the attached emails for a full explanation on how to obtain up to 750GH out of those boards.

Logistics:

Once the group buy is arranged the ASICs take about a week to be tested and shipped to the chinese board assembly shop. The chinese supplier HashFast is using promised to assemble and ship the first 1,500 boards within 18 days from the reception of the order and parts. That means that this thread needs to collect enough preorders to make the batch worth it’s run, wait for the chip testing and shipping, the board supplier 18 days and then for the delivery courier. Hashfast is waiting for confirmation on the lead times for the bolster plates, cooling pads and indium foil. It is expected to be less than the lead time for the boards. The kit can be operated without the bolster plate/cooling pad/foil at a reduced performance.

---

Original emails:

Initial email:

Hi Davide,
One of our China manufacturers has just offered us very good pricing on their remaining stock of rev3 boards. This could potentially enable a rev3 group buy. We can supply them with the long lead time parts, and the ASIC, they can produce the board, and we can supply a cooler. [...] Boards ship from China, coolers from US or Canada. The first 1500 boards can be ready 18 days after order placed according to the manufacturer. The China manufacturer has asked for a minimum order of 2000.
Simon

Details about board performances:

Yes, seems OK. I think it is also very possible that the manufacturer will accept a slightly smaller order, if offered. We need to coordinate how the logistics will work. [...].
In general getting something out there sooner rather than later would be good to gauge interest while we work out the logistical details.
With a bolster plate and indium foil (not included in what I've talked about so far) the boards can do 700GH and with careful setup 750GH can be achieved. Note - these numbers are based on our previous chip testing regime - the higher volume chip testing we are doing now could allow a small number of lower performing chips to be passed.

More about board performances:

I've double checked with one of the engineers about speeds, before publishing anything. [...] For the basic board / cooler combination with good thermal grease without bolster plate, he thinks that it's easy to get 600MHz, and with a little skill 700MHz. Higher speeds are possible, but you need to be very careful in how gently and evenly you tighten the cooler. This corresponds to theoretical speeds of 461GH/s or 538GH/s. I'm going to double check these numbers with our China manufacturer, who has done their own extensive testing too. To get higher speeds requires a bolster plate, so the board does not distort and warp when you tighten the cooler down. For highest speeds you need indium foil instead of grease between the chip and the cooler. I'm double checking the numbers for the boards with bolster plate and foil, but historically 700GH has been easy and 750GH with careful setup. Bolster plates are not expensive to get made, but would complicate the project. (You need the bolster plate, a piece of thermal pad to go over it, and a piece of indium foil between the chip and the cooler).

Efficiency at lower hashrate:

In order to set up the boards without the bolster plate, and still achieve good thermal mounting, our China assembly partner uses jig with a clamp that evenly holds the cooler down onto the board, and applies a good pressure, and then the assembly worker screws the cooler on, just turning the screws hand tight. This jig allows the unskilled assembly worker to achieve good, consistent, even mounting pressure.

The same can be achieved by hand, as long as you are careful to do the tightening evenly. One of the software setup tools loads up the chip with dummy work, displays a continuous readout of die temperature so you can watch for even thermal contact as you tighten up the screws.

BTW The china manufacturer reports that running the boards at 450GH he measures 0.8J/GH at the wall.

Components required to push the board to 750GH:

The bolster plate is custom for the board. I don't know of anything off the shelf that would work. They are not expensive, but I will need to get a quote to ensure lead time and price. We have already had a fair number made, so this would be a simple repeat order. A piece of thermal pad is needed between the bolster plate and the board to prevent the metal plate from shorting the components on the bottom of the board. These come in 20x40cm sheets from a chinese manufacturer, and need to be cut into squares with scissors. Thirdly the indium foil comes from the Indium corporation, I will need to check if we have any stock left, or lead time to order.

About logistics:

The biggest question is logistics - if all of this goes to a single buyer, who can do the final assembly, then it's relatively easy. If many small buyers are involved then it's more work.

No problem - I can arrange the ordering of bolster plates/pads/foil.

Our China manufacturing partner will ship the boards to multiple customers though. They need a shipping solution, and I have offered to send them the design for the shipping solution we used to ship boards to Ciara in Montreal. This solution is a specially designed cardboard box that takes 30 boards, and offered good protection for them.

The China partner is using the same coolers as us, but got better performance numbers than our engineer reported for the stock support/grease configuration (which we did not experiment with much - we focussed on the enhanced bolster plate/foil). They got 550GH from 98% of their boards, and 600GH from about 70% of the boards. This is at 35C ambient, so would do better in a cooler place. Given this variability I think we need to offer the boards at a fixed price per board, and let people know the different performance range that can be expected, so they can calculate $/GH themselves. In addition they stated that, like us, the exact process of mounting of the cooler to the board is critical. Applying the pressure evenly and gently is important. Doing it well makes a big difference in performance, and since we'd be leaving this to the end users we can't guarantee what their work will be like, so we can't make any statement about performance.

Shipping details:

Getting the shipping boxes made for the boards will not cost much, but shipping will have to be insured, to make sure any damage loss is covered. These prices won't allow much buffer for handling returns.

We would ship coolers from the USA or Canada in boxes of 4, and fans in boxes of 20. The end user would need to purchase a PSU, screws, and standoffs to complete their rig.

Also - given that the coolers come in boxes of 4, the fans in boxes of 20, and the shipping boxes for boards would typically carry 30 or so (but can be made any size), I think it might be difficult to support customers ordering 1 or 2 - since it will require a lot of repacking. The whole point of a group buy is to reduce the number of orders (ideally to 1) so that the consolidated shipping/handling makes all these things easier/cheaper. Perhaps an ordering unit of 20 boards would work - although more would be better. The China manufacturer can have the board shipping boxes made up to carry 20 boards, and we can ship 5 boxes of 4 coolers and 1 box of fans to each customer. We will also need some packaging (a box) to ship the 20 coolers, 20 fans, 20 bolster plates, thermal pad material for 20 boards, and 20 pieces of foil.