Swedish ASIC miner company kncminer.com

[ASIC-K]

Does it concern anyone that Bitfury has chips in the field being tested with images and videos available?
Will we see similar from KNC? Aren't they both projecting miner availability in September?
https://bitcentury.io/blog/bitfury-asic-in-action

Bitfury got just sample chips 5gh on video only 1,2gh passive cooling, they even do not now what happed with chips at 5gh with cooling. The knc if it's true 100gh per chip doesn't need so many per unit.
I know it's stupid question but is pre bordering ope
n or closed 4 mars?

Except that the KNC chips are MASSIVE in size (expensive to produce) and the Bitfury chips are TINY (and thereby CHEAP).

KNC - ~3000mm2
BitFury - ~16mm2

I've gone over this before about why HUGE chips are very very bad for profitability (i.e. low chip yields)
https://bitcointalk.org/index.php?topic=244584.msg2680796#msg2680796


In full disclosure, I have a Bitfury order in process.  I bought it because that was looking to be first to market when I made my purchase.  I also have Terrahash on order, and that's looking to be a painful experience.

i still dont understand why they went with such a giant die size? clearly i am no engineer, but i thought 28nm was supposed to be SMALLER.

[1714034234]

1714034234

[1714034234]

1714034234

[bkpduke]

i still dont understand why they went with such a giant die size? clearly i am no engineer, but i thought 28nm was supposed to be SMALLER.

28nm is simply the lithography line width.  It does allow you to pack more transistors into the same space.

The problem is their design, IMHO.  They are trying to produce a "super chip" with many many many SHA256 engines in it.  Other companies are making fewer engines, and "uniting" them on the board.

The theoretical of what KNC is doing is better, but the practical (i.e. what we know from 30 years of IC production) means it is going to have a high rate of defective or partially defective chips.

It also means it will be a HOT chip and take some good cooling (probably why they have designed things in a "wind tunnel" type design).

[ASIC-K]

i still dont understand why they went with such a giant die size? clearly i am no engineer, but i thought 28nm was supposed to be SMALLER.

28nm is simply the lithography line width.  It does allow you to pack more transistors into the same space.

The problem is their design, IMHO.  They are trying to produce a "super chip" with many many many SHA256 engines in it.  Other companies are making fewer engines, and "uniting" them on the board.

The theoretical of what KNC is doing is better, but the practical (i.e. what we know from 30 years of IC production) means it is going to have a high rate of defective or partially defective chips.

It also means it will be a HOT chip and take some good cooling (probably why they have designed things in a "wind tunnel" type design).

man, thats the best explanation i have heard so far. thanks for that!!

[erk]

Except that the KNC chips are MASSIVE in size (expensive to produce) and the Bitfury chips are TINY (and thereby CHEAP).

KNC - ~3000mm2
BitFury - ~16mm2

I've gone over this before about why HUGE chips are very very bad for profitability (i.e. low chip yields)
https://bitcointalk.org/index.php?topic=244584.msg2680796#msg2680796


In full disclosure, I have a Bitfury order in process.  I bought it because that was looking to be first to market when I made my purchase.  I also have Terrahash on order, and that's looking to be a painful experience.

Nonsense, their package is 3,000mm2 not the die size.

[bkpduke]

Except that the KNC chips are MASSIVE in size (expensive to produce) and the Bitfury chips are TINY (and thereby CHEAP).

KNC - ~3000mm2
BitFury - ~16mm2

I've gone over this before about why HUGE chips are very very bad for profitability (i.e. low chip yields)
https://bitcointalk.org/index.php?topic=244584.msg2680796#msg2680796


In full disclosure, I have a Bitfury order in process.  I bought it because that was looking to be first to market when I made my purchase.  I also have Terrahash on order, and that's looking to be a painful experience.

Nonsense, their package is 3,000mm2 not the die size.

Link for proof please.  You don't make a package size of 3000mm2 if you have tiny chips . . .

[dwdoc]

I cannot believe no one else has picked up on this as well.  KNCMiner's die size is 2.5 INCHES squared.  That's . . . . Titanic sized.

I hate do to be debbie downer here, but that means that on a standard 300mm wafer they will not get many chips, and that's assuming a 100% chip yield from the wafer.  Wafer's always have defects which cause some chips to go down.  The yield decreases exponentially as a the die size goes up (i.e. it only takes a tiny imperfection in the wafer to make the entire chip a paperweight).

Even if they get chips, the yield is going to be absurdly low, I would bet less than 25%.  They can probably recoup some investment by selling chips that are spec'd lower and have some "cores" disabled in them, but this is hit or miss depending on where in the chip the defect occurs.

Of course we picked up on it. But, OrSoc are not beginners and my understanding is the chips have an almost gpu-like design (massively parallel). Therefore , if a few units do not work, they will be disabled upon self-test. Also, the chips have extra units to compensate for defects in the manufacturing process. Also, my understanding was that they over-engineered this chip in order to account for as many issues as they possibly could.

Now this is just my impression from reading the Q/A session and watching the videos, but it seems to me that they are taking a (very) calculated risk. Even if a chip is massive, the fact that is not very complicated should probably increase yields, so we'll just have to wait and see.

There is a reason companies like Intel do not make these MASSIVE chips.  No one is as good as Intel in chip manufacturing, not even close.  If Intel actively avoids this die sizes, there is a @#$N good reason.

This is no my first rodeo, I follow IC manufacturing closely, and have for 20 years now.  Making a chip this big is a huge huge mistake.  If they are even making it at all (where are the die shots?).


EDIT - and chip yields have just as much to do about wafer quality as the IC design.  The bigger these things get, the faster yields fall.  Even if they are like a GPU and can shut off parts of the chip, there is no way around that.

I think we'll all feel better when we see some chip die shots. Hopefully soon...

[erk]

Link for proof please.  You don't make a package size of 3000mm2 if you have tiny chips . . .

More nonsense, Pentium III was 49.5 x 49.5mm, Pentium 4 53.3mm x 53.3mm  so 55 x 55mm is hardly unusual considering you are trying to dissipate twice the heat of those old processors.

[Fiyasko]

Does it concern anyone that Bitfury has chips in the field being tested with images and videos available?
Will we see similar from KNC? Aren't they both projecting miner availability in September?

https://bitcentury.io/blog/bitfury-asic-in-action

OOO! OO!, THAT LOOKS LIKE WHAT I WANT, WHERE CAN I BUY THAT!

[erk]

Does it concern anyone that Bitfury has chips in the field being tested with images and videos available?
Will we see similar from KNC? Aren't they both projecting miner availability in September?

https://bitcentury.io/blog/bitfury-asic-in-action

OOO! OO!, THAT LOOKS LIKE WHAT I WANT, WHERE CAN I BUY THAT!

Try their thread not here please.

[Fiyasko]

Does it concern anyone that Bitfury has chips in the field being tested with images and videos available?
Will we see similar from KNC? Aren't they both projecting miner availability in September?

https://bitcentury.io/blog/bitfury-asic-in-action

OOO! OO!, THAT LOOKS LIKE WHAT I WANT, WHERE CAN I BUY THAT!

Try their thread not here please.

They have a thread?, all i could find was a thread about the bitfury Chips themsleves, I want to purchase a Plug-Set-Go ASIC

[Templer]

Does it concern anyone that Bitfury has chips in the field being tested with images and videos available?
Will we see similar from KNC? Aren't they both projecting miner availability in September?

https://bitcentury.io/blog/bitfury-asic-in-action

OOO! OO!, THAT LOOKS LIKE WHAT I WANT, WHERE CAN I BUY THAT!

Try their thread not here please.

They have a thread?, all i could find was a thread about the bitfury Chips themsleves, I want to purchase a Plug-Set-Go ASIC

Please this is a KNC Thread!!

[dwdoc]

Does it concern anyone that Bitfury has chips in the field being tested with images and videos available?
Will we see similar from KNC? Aren't they both projecting miner availability in September?

https://bitcentury.io/blog/bitfury-asic-in-action

OOO! OO!, THAT LOOKS LIKE WHAT I WANT, WHERE CAN I BUY THAT!

Try their thread not here please.

They have a thread?, all i could find was a thread about the bitfury Chips themsleves, I want to purchase a Plug-Set-Go ASIC

https://bitcointalk.org/index.php?topic=251966.0

[jbutter]

No need for the ATX PWS if you chose Hosting, and if the unit breaks, no RMA headaches, plus they mentioned they tweak it to perfection.

[Mota]

well, I find 350$ hard to swallow.

[de_ixie]

well, I find 350$ hard to swallow.

Will be highly dependend on your location and what you pay for energy. I calculated it earlier in this thread. In Germany for instance you pay on average 0.25€/ 0.32$ for 1kW/h.

This means around 230$ per month electricity for a Jupiter. Adding all the hosting-services they will (probably) provide 350$ is a pretty attractive offer. In areas with lower power costs this may not be the case.

[Mota]

yes, but there is the 2% fee... For me it was either one or the other, but certainly not both. I was willing to pay up to 4% in hosting fees, but a dollar fee and a percentage fee is too much imho...

[jbutter]

Don't forget the cooling fees. I'm sure the pool will be in the same hosting center with low latency, but it should be an added value (0%).

[bkpduke]

Link for proof please.  You don't make a package size of 3000mm2 if you have tiny chips . . .

More nonsense, Pentium III was 49.5 x 49.5mm, Pentium 4 53.3mm x 53.3mm  so 55 x 55mm is hardly unusual considering you are trying to dissipate twice the heat of those old processors.

Having math difficulty?

The Pentium III had a die size of 80mm2 (Coppermine version):
http://ark.intel.com/products/27531/Intel-Pentium-III-Processor-1_13-GHz-256K-Cache-133-MHz-FSB
That's 8mm x 10mm.  A fraction of the size you thought it was.

The Pentium 4 die size was 131mm2
http://ark.intel.com/products/27438/Intel-Pentium-4-Processor-2_40-GHz-512K-Cache-533-MHz-FSB
That's about 11mm x 11.9mm  Also a fraction of the size you thought it was.

55 x 55mm is 3025mm2
- 37.8 times the die size of the Pentium 3
- 23.1 times the die size of the Pentium 4

Now, I can do this all day long, so please continue.  

[bitpop]

Yes please, their chips are as big as toasts! Mmmmm

[bitpop]

Thank you. Please refer to them as toasts.
I would love to buy broken ones!