CHINA -Is there Companies/People Who manufactures the ASICs Chips for BTC mining

[sajidfbi]

Hi,
 Is there any Companies/People from CHINA who are manufacturing or can manufactures ASIC chips, Is there any one who can create 5Th upto 10Th Mining Machines ? I am looking for some bulk provider or creator, i don't want to go ahead with some available expensive provider such as Antminer or Spondolies, My Thread is about China so if there's people from china who can create BTC mining machines at low cost let me know Please. Or any help from any expert ? i can even visit China personally in this matter if necessary. 

[philipma1957]

This post seems like you are putting a sign on yourself.  


Rob me please Rob me.


A) do you have a person you trust that speaks Chinese?
B) if you do ask him why I think your idea will get you robbed.

[sajidfbi]

This post seems like you are putting a sign on yourself.  


Rob me please Rob me.


A) do you have a person you trust that speaks Chinese?
B) if you do ask him why I think your idea will get you robbed.

I think it's not about robbery, and I am not fool to pay out some one blindly, I have already mentioned in my post if necessary I might personally go to China.

[innerchaos]

why go to china just order a Smart Miner they are beast

[Finksy]

As I mentioned in your other thread, the cost of chips makes up upwards of 70-80% of the total cost to produce a miner.  All the manufacturers that produce their own chips also produce their own miners, as the chips are the important, technologically significant and expensive part.  If there were cost savings to be had by going closer to the source, the manufactures themselves would be taking them.  If you have to buy the chips second hand from one of the hardware manufacturers, they will charge a premium, and since that cost makes up a great majority of the total cost of the miner, you are unlikely to save much money by producing the miner yourself.  Unfortunately, it's an industry of scale.  And with mining being a fiercely competitive market, and most of these manufacturers self-mining as well as selling, any cost margins that can be reduced typically are already, so unless you are capable of developing your own chips, or somehow making existing miners more efficient (like what Bitmain did by going with a 2-chip string design), it's unlikely that you will be able to save any significant amount of money by building your own miners. Now, if you are talking about buying chips in very large quantity (think millions), it's possible you could get a good enough price to save money over buying existing miners.  Contact the manufacturers like Bitmain, Spondoolies, Bitfury, Sfards, Avalon and see if they're willing to sell chips to you in the volume you are looking at.

[Finksy]

why go to china just order a Smart Miner they are beast

Can you show me anybody who has received and evaluated one of these miners? They are claiming 0.105 kw/GHs, which has not yet been done by any of the large, reputable manufacturers.  I'm leaving negative trust on your account until you can show evidence that this miner exists or you weren't serious, otherwise you are clearly involved with aiding in a scam.

[sidehack]

Seems more likely that he's being sarcastic, offering you an obviously terrible solution as a joke.

[Finksy]

Maybe, I can be pretty dense when it comes to sarcasm online at times.

BTW what's your take on the question at hand sidehack? You would know at least as well as anyone whether it would be worth while buying chips second hand and developing a large, efficient stand-alone miner.  Are the cost savings really there if you have to buy the chips from the manufacturers?

[sidehack]

Novak and I estimated we could build a BM1384-based miner getting a similar efficiency to the S5 at the top end (2400W for about 5TH) and pretty good around the bottom end (2.5TH about 650W) for around $1200. That's an envelope estimate without a thorough design evaluation and doesn't really include markups for labor and such, but I think included PSU. A fixed-setpoint miner like the S5 would be a bit cheaper (and marginally more efficient at the top end), but the added cost of making it a lot more variable greatly extends the effective service life and increases total profitability.

Bitmain designs are fairly simple and straightforward, so probably not that easy to underbid unless you can actually do the manufacturing yourself. Spondoolies will be difficult to beat on density (if you care about that) but really easy to beat on price and complexity.

[innerchaos]

why go to china just order a Smart Miner they are beast

Can you show me anybody who has received and evaluated one of these miners? They are claiming 0.105 kw/GHs, which has not yet been done by any of the large, reputable manufacturers.  I'm leaving negative trust on your account until you can show evidence that this miner exists or you weren't serious, otherwise you are clearly involved with aiding in a scam.

Oh its my responsibility to ensure you don't get ripped off.. or that the miner exists..   when did I sign up for that task?

I googled 20th miner and google spit out a result and it was the TOP #1 Hit.. Google cant possibly lie. they are worth 250 Billion dollars. if you cant trust a 250 Billion dollar company then what is this world coming to.

[innerchaos]

Seems more likely that he's being sarcastic, offering you an obviously terrible solution as a joke.

And yes of course it was a joke... thank you for seeing it as it was meant... Smart Miners have been the butt of many jokes on this and other forums... and most readers already know of them.

[Finksy]

Oh its my responsibility to ensure you don't get ripped off.. or that the miner exists..   when did I sign up for that task?

I googled 20th miner and google spit out a result and it was the TOP #1 Hit.. Google cant possibly lie. they are worth 250 Billion dollars. if you cant trust a 250 Billion dollar company then what is this world coming to.

I know that it's a scam, the point was when members on here actively try to promote scams, they get labeled with negative trust.  It's not your responsibility to do anything you don't want, but many people on this forum try to help less experienced people from getting ripped off. There was nothing in your post that denoted it as being fictitious, don't take it so personally.

Also your lack of punctuation and piss poor writing was very consistent with the typical scammers we see around here.

[philipma1957]

Seems more likely that he's being sarcastic, offering you an obviously terrible solution as a joke.

And yes of course it was a joke... thank you for seeing it as it was meant... Smart Miners have been the butt of many jokes on this and other forums... and most readers already know of them.

got here too late to suggest you were kidding.

@op  

right now there is one chip that you may be able to buy it is the chip  in bitmaintech's  s-5

and there is only one guy I know that can use a chip from that miner to build  a better miner.

Sidehack, has shown he can do this.   and on the smallest scale a 1 chip 10gh miner

____________________________________________________________
Op you need understand the club is very closed at this time.

real gear:

avalon 4.1 sold out---------------China

sp-tech sp20 sold out------------Israel

bitmaintech s-5 still for sale----China

bitfury does not sell--------------Some say Russian based
knc does not sell------------------Sweden


maybe Sfards new gear works.-----China


I should have been more clear in my thread.

Do you have  a trusted Chinese friend?
Do you have a trusted friend with Chinese contacts.

If you do ask them to help you.  Post in the Chinese Forum section.


https://bitcointalk.org/index.php?board=30.0


The best guy I do business with in China hosts miners maybe he can broker a deal for you 40 or 50 used s-5's.


This is his thread.


https://bitcointalk.org/index.php?topic=934581.0

here is his profile link:

https://bitcointalk.org/index.php?action=profile;u=253683

ask if he will procure a bulk order for you with psu's  

maybe he can save money for you.

He has been pretty honest sending him a pm and asking on his thread may be a good move for you.

[HyperMega]

As I mentioned in your other thread, the cost of chips makes up upwards of 70-80% of the total cost to produce a miner.

For a strict string design without DC/DCs I guess that the pure ASIC chip costs are even more than 90% of the miner production costs. So you can only optimize the remaining 10% by going to China for the final miner assembly.

Bitmain, SFARDS, SP, KNC and the like do not produce their chips. They pobably don't even have designed and implemented them completely.
What they did is to place an order at an ASIC design house for implementing their chip architecture in a technology of one of the few foundries (TSMC, GF, UMC, Samsung, Intel, STM, SMIC) providing advanced nodes (28nm Bulk/28nm FDSOI, 22/20nm, 16/14nm), which finally manufacture the wafers containing their chips.

Having this in mind, I think we reached a kind of a sweet spot with the highly optimized 28nm designs.
TSMC is the largest foundry and a public traded company. They published their revenues with 28nm wafers. Based on that one can calculate that the average selling price of a wafer if about $6k USD. For less than a billion deals like a Bitcoin ASICs >$8k USD per wafer is more realistic.

Assuming 80 TH/s per 28nm wafer (depends on chip design quality/yield and frequency@voltage operation point) one would already have about $0.10 USD per GH/s production costs for the pure 28nm silicon.
Add $0.02 USD per GH/s for packaging (very rough estimation based on a QFN package and 24 GH/s per package).
And if you want to recover about $2M USD NRE for design services and mask set you have to add additional $0.02 USD (based on 100 PH/s sales).

Finally you will end up with production costs of about $0.14 USD per GH/s for 28nm ASIC.

Bitmain currently sells complete miners at $0.31 GH/s. So there is not so much room for a price war.

You may ask, what about 16/14nm? Of course
they could be more engery efficient (still waiting for some real KNC measurements), but there is currently no way to produce them cheaper as 28nm ASICs.

[sajidfbi]

As I mentioned in your other thread, the cost of chips makes up upwards of 70-80% of the total cost to produce a miner.

For a strict string design without DC/DCs I guess that the pure ASIC chip costs are even more than 90% of the miner production costs. So you can only optimize the remaining 10% by going to China for the final miner assembly.

Bitmain, SFARDS, SP, KNC and the like do not produce their chips. They pobably don't even have designed and implemented them completely.
What they did is to place an order at an ASIC design house for implementing their chip architecture in a technology of one of the few foundries (TSMC, GF, UMC, Samsung, Intel, STM, SMIC) providing advanced nodes (28nm Bulk/28nm FDSOI, 22/20nm, 16/14nm), which finally manufacture the wafers containing their chips.

Having this in mind, I think we reached a kind of a sweet spot with the highly optimized 28nm designs.
TSMC is the largest foundry and a public traded company. They published their revenues with 28nm wafers. Based on that one can calculate that the average selling price of a wafer if about $6k USD. For less than a billion deals like a Bitcoin ASICs >$8k USD per wafer is more realistic.

Assuming 80 TH/s per 28nm wafer (depends on chip design quality/yield and frequency@voltage operation point) one would already have about $0.10 USD per GH/s production costs for the pure 28nm silicon.
Add $0.02 USD per GH/s for packaging (very rough estimation based on a QFN package and 24 GH/s per package).
And if you want to recover about $2M USD NRE for design services and mask set you have to add additional $0.02 USD (based on 100 PH/s sales).

Finally you will end up with production costs of about $0.14 USD per GH/s for 28nm ASIC.

Bitmain currently sells complete miners at $0.31 GH/s. So there is not so much room for a price war.

You may ask, what about 16/14nm? Of course
they could be more engery efficient (still waiting for some real KNC measurements), but there is currently no way to produce them cheaper as 28nm ASICs.

Hi HyperMega, So far you have posted a very helpful review, you seems very experience with Asic's etc, As per you said above that

(Finally you will end up with production costs of about $0.14 USD per GH/s for 28nm ASIC.
Bitmain currently sells complete miners at $0.31 GH/s. )

So i think there's still a very Huge Difference in price, I have purchased few S5 from Antminer with Local China Shipping, It cost me $410 each S5 which is about 0.36 per GH/S, So i think if there's a possible way to create a Miner with about $0.14 Per Gh/s then it's mean we can get an Miner of 1125 GH/s at $157 approximately instead of $410, And if we go for like 100 Machines (Miners) with such price then we can save up to $25300, I think it's good enough to save such amount.

I am just going to send you a PM, Please have a look.

[spazzdla]

why go to china just order a Smart Miner they are beast

SCAM SCAM SCAM SCAM

DO NOT BUY THIS IS A SCAM

YOU HAVE BEEN WARNED.

[sidehack]

Getting a chip manufactured is looking at $0.14/GH which is about right based on the numbers I've gotten from Bitmain for bulk sales. Building a miner out of those chips for zero cost is terribly unfeasible, especially if you only want 100 - economies of scale will greatly reduce per-unit costs as the number of units increases.

[sajidfbi]

Getting a chip manufactured is looking at $0.14/GH which is about right based on the numbers I've gotten from Bitmain for bulk sales. Building a miner out of those chips for zero cost is terribly unfeasible, especially if you only want 100 - economies of scale will greatly reduce per-unit costs as the number of units increases.

by 100 i mean it was just a suppose quantity for showing the difference, i would definitely like to create atleast 3 to 5 Ths machines with those chips as i don't have problem of electricity, so instead of creating 1000 machines each with 1 THs i would prefer to create 100 with 3 upto 5 THs each.

[HyperMega]

Getting a chip manufactured is looking at $0.14/GH which is about right based on the numbers I've gotten from Bitmain for bulk sales. Building a miner out of those chips for zero cost is terribly unfeasible, especially if you only want 100 - economies of scale will greatly reduce per-unit costs as the number of units increases.

Intressting! Bitmain offered you chips for about $0.14/GH for bulk sales? Then my estimations may be wrong, because they also would not sell chips in bulk without profit. 

In general I think a margin up to 100% on hardware sales is ok, because the hardware manufacturer has to provide support & warranty and they also must finance the developent of their next generation.
Just remember for a moment the wild times in 2013, when 28nm hashing power was sold for $10/GH. More than 1000% margin for the hardware manufacturer!

Have you ever stripped one of the latest Bitmain chips to have a look at the silicon die?
Based on the number of hashing cores I would estimated it should be between 18mm2 ... 20mm2. Can you confirm this?

[sidehack]

Yeah, I bought into Sushi's second round of sales on the S1 which set me back something like $4k. It's crazy to me that I had that kind of money in coin at the time just from mining, after buying a couple AM Cubes for almost $3k a month earlier. Stuff was all kinds of expensive then. I'd probably have sold out quickly if I was building boards back then on account of you'd be hard-pressed to talk us into a 50% markup, let alone 100-1000.

I have not stripped a chip.