KNCMiner and their 'magic' SHA256 alogorithm

[Bitcoinorama]

OK, 2 problems:

1/ they can't get a better/other algo, it's SHA256 or nothing.

2/ they're a joke on the BUSINESS side. (relates to false claim about 1/).

They're full of it, period. This is not about the SHA256 algo, it's about their continuous BS. ORSoC is not making those stupid claims, it's all KNCMINER's doing (which brings me again to the fact that ORSoC is NOT KNCMINER, so don't confuse the two, even if they are in close ties - it'll be KNCMINER losing your money, if).

Get the head OUT of the tech, it's only a small part of it. It's the business that will screw you (if).

You obviously are not a software engineer. First, they never claimed to have improved the sha256 algorithm. You said that, not them. There is more to coding a usable ASIC than just the Algorithm, and there is always more than one way to code software. Saying that they improved their algorithm does not imply that they were referring to improving the SHA256 algorithm. They may have just been referring to their implimentation of it, or the code they wrote to support it. For example, if i code my asic to run 4 instances or SHA256, and then realize that there was actually room for 5. i could recode it to run 5. I've improved my algorithm without improving SHA256. I think this is a case of you not understanding what they meant.

You're right, I jumped the gun on the "SHA256" bit.

Here's their website quote:

"An additional gain of 30% more hashing when the advanced algorithms provided by ORSoC are applied."

30% more hashing than what then?

Mars = 5.1 Gh/s stock. Mars at the open day = 6.8 Gh/s.

30% more hashing power = 6.63 Gh/s

Saturn = 125 Gh/s stock.

Jupiter = 250 Gh/s stock.

You do the math...

[markj113]

Shouldn't that be 175 / 350 stock

[Bitcoinorama]

Shouldn't that be 175 / 350 stock

No, it was 250 Gh/s originally advertised, and they said they intended to over deliver. Then they said it was with a +30% improvement at 350 G/hs, and were introducing a mid-level price entry as they were splitting Jupiter into two units. Saturn was then born at 175 Gh/s, which one assume is 125 Gh/s stock, if Jupiter is 250 Gh/s stock. Mars was proof of concept which was why one of my first questions on the open day was, what is Mars' stock speed, is it 6 and you intend to OC, hence is the 6.8 part of the way there? but it was the case that Mars is 5.1 Gh/s in stock FPGA, and the coding has clocked it at 6.8 Gh/s stable Monday week back. Marcus aims to push it further currently, their improvements have more to yield. I assume any improvements now make it on further ASIC revisions in future, not this time round, but I maybe wrong...

[blastbob]

Shouldn't that be 175 / 350 stock

No, it was 250 Gh/s originally advertised, and they said they intended to over deliver. Then they said it was with a +30% improvement at 350 G/hs, and were introducing a mid-level price entry as they were splitting Jupiter into two units. Saturn was then born at 175 Gh/s, which one assume is 125 Gh/s stock, if Jupiter is 250 Gh/s stock. Mars was proof of concept which was why one of my first questions on the open day was, what is Mars' stock speed, is it 6 and you intend to OC, hence is the 6.8 part of the way there? but it was the case that Mars is 5.1 Gh/s in stock FPGA, and the coding has clocked it at 6.8 Gh/s stable Monday week back. Marcus aims to push it further currently, their improvements have more to yield. I assume any improvements now make it on further ASIC revisions in future, not this time round, but I maybe wrong...

It has been advertised with 350GHs with possible 30% more with optimization..

[Bitcoinorama]

Shouldn't that be 175 / 350 stock

No, it was 250 Gh/s originally advertised, and they said they intended to over deliver. Then they said it was with a +30% improvement at 350 G/hs, and were introducing a mid-level price entry as they were splitting Jupiter into two units. Saturn was then born at 175 Gh/s, which one assume is 125 Gh/s stock, if Jupiter is 250 Gh/s stock. Mars was proof of concept which was why one of my first questions on the open day was, what is Mars' stock speed, is it 6 and you intend to OC, hence is the 6.8 part of the way there? but it was the case that Mars is 5.1 Gh/s in stock FPGA, and the coding has clocked it at 6.8 Gh/s stable Monday week back. Marcus aims to push it further currently, their improvements have more to yield. I assume any improvements now make it on further ASIC revisions in future, not this time round, but I maybe wrong...

It has been advertised with 350GHs with possible 30% more with optimization..

No it hasn't.

[blastbob]

Must have misread, anyway i have not promised anyone more than 350

[idee2013]

Shouldn't that be 175 / 350 stock

No, it was 250 Gh/s originally advertised, and they said they intended to over deliver. Then they said it was with a +30% improvement at 350 G/hs, and were introducing a mid-level price entry as they were splitting Jupiter into two units. Saturn was then born at 175 Gh/s, which one assume is 125 Gh/s stock, if Jupiter is 250 Gh/s stock. Mars was proof of concept which was why one of my first questions on the open day was, what is Mars' stock speed, is it 6 and you intend to OC, hence is the 6.8 part of the way there? but it was the case that Mars is 5.1 Gh/s in stock FPGA, and the coding has clocked it at 6.8 Gh/s stable Monday week back. Marcus aims to push it further currently, their improvements have more to yield. I assume any improvements now make it on further ASIC revisions in future, not this time round, but I maybe wrong...

It has been advertised with 350GHs with possible 30% more with optimization..

sorry...but:

 +1

and this time KS is right, too:

Here's their website quote:

"An additional gain of 30% more hashing when the advanced algorithms provided by ORSoC are applied."

30% more hashing than what then?

have a look https://www.kncminer.com/products/jupiter

[titomane]

OK, 2 problems:

1/ they can't get a better/other algo, it's SHA256 or nothing.

2/ they're a joke on the BUSINESS side. (relates to false claim about 1/).

They're full of it, period. This is not about the SHA256 algo, it's about their continuous BS. ORSoC is not making those stupid claims, it's all KNCMINER's doing (which brings me again to the fact that ORSoC is NOT KNCMINER, so don't confuse the two, even if they are in close ties - it'll be KNCMINER losing your money, if).

Get the head OUT of the tech, it's only a small part of it. It's the business that will screw you (if).

You obviously are not a software engineer. First, they never claimed to have improved the sha256 algorithm. You said that, not them. There is more to coding a usable ASIC than just the Algorithm, and there is always more than one way to code software. Saying that they improved their algorithm does not imply that they were referring to improving the SHA256 algorithm. They may have just been referring to their implimentation of it, or the code they wrote to support it. For example, if i code my asic to run 4 instances or SHA256, and then realize that there was actually room for 5. i could recode it to run 5. I've improved my algorithm without improving SHA256. I think this is a case of you not understanding what they meant.

You're right, I jumped the gun on the "SHA256" bit.

Here's their website quote:

"An additional gain of 30% more hashing when the advanced algorithms provided by ORSoC are applied."

30% more hashing than what then?

Mars = 5.1 Gh/s stock. Mars at the open day = 6.8 Gh/s.

30% more hashing power = 6.63 Gh/s

Saturn = 125 Gh/s stock.

Jupiter = 250 Gh/s stock.

You do the math...

125 + 30% is 162.5
250 + 30% is 325

[Bitcoinorama]

OK, 2 problems:

1/ they can't get a better/other algo, it's SHA256 or nothing.

2/ they're a joke on the BUSINESS side. (relates to false claim about 1/).

They're full of it, period. This is not about the SHA256 algo, it's about their continuous BS. ORSoC is not making those stupid claims, it's all KNCMINER's doing (which brings me again to the fact that ORSoC is NOT KNCMINER, so don't confuse the two, even if they are in close ties - it'll be KNCMINER losing your money, if).

Get the head OUT of the tech, it's only a small part of it. It's the business that will screw you (if).

You obviously are not a software engineer. First, they never claimed to have improved the sha256 algorithm. You said that, not them. There is more to coding a usable ASIC than just the Algorithm, and there is always more than one way to code software. Saying that they improved their algorithm does not imply that they were referring to improving the SHA256 algorithm. They may have just been referring to their implimentation of it, or the code they wrote to support it. For example, if i code my asic to run 4 instances or SHA256, and then realize that there was actually room for 5. i could recode it to run 5. I've improved my algorithm without improving SHA256. I think this is a case of you not understanding what they meant.

You're right, I jumped the gun on the "SHA256" bit.

Here's their website quote:

"An additional gain of 30% more hashing when the advanced algorithms provided by ORSoC are applied."

30% more hashing than what then?

Mars = 5.1 Gh/s stock. Mars at the open day = 6.8 Gh/s.

30% more hashing power = 6.63 Gh/s

Saturn = 125 Gh/s stock.

Jupiter = 250 Gh/s stock.

You do the math...

125 + 30% is 162.5
250 + 30% is 325

lol. I didn't actually mean you do th...

[WillMilk4Coin]

Shouldn't that be 175 / 350 stock

No, it was 250 Gh/s originally advertised, "....."

It has been advertised with 350GHs with possible 30% more with optimization "...."

sorry...but:
 +1
and this time KS is right, "...."
have a look https://www.kncminer.com/products/jupiter

Correct. There is only one possible conclusion that can be drawn from the listings in their current form, namely, at stock speeds Saturn and Jupiter run at 175 and 350 Gh, respectively. Furthermore, semantically speaking, customers should expect a 30% improvement upon the given figures as well. In other words, customers can expect Saturn to put out ~227 Gh/s and Jupiter, 455 Gh/s. Of course, the safest position for a prospective customer is to take only the listed hash speeds at face value and assume no more, no less.

If the 30% improvement is indeed included into the given hash speeds for both devices, then that 30% statement on the product listing is not only redundant, but extremely misleading.

[Trongersoll]

and this is the first time people with English as a second language have said something not quite right. heck, JFK said in german that he was a jelly doughnut.

[titomane]

Marcus told me that their products worked without factory OC. What then if the client wanted to OC, at your own risk.

[WillMilk4Coin]

and this is the first time people with English as a second language have said something not quite right. heck, JFK said in german that he was a jelly doughnut.

Was your sarcastic remark in response to my statement? I will not argue what may or may not have been lost in translation on part of KnC when they made their listings. I was simply pointing out there isn't any other way to interpret what is being stated on their product page and understanding it to mean anything else at all is assumption. Let's say the 30% increase is already applied in the 350 Gh spec given for Jupiter, then pointing it out later would be entirely irrelevant as well, would it not? To further reinforce the idea that the 30% is, in fact, intended to be applied in addition to the given figures, they also use the quantifier, "minimum", before the units' speeds. Lastly, if memory serves me correctly, Sam Cole is largely responsible for their website. Sam Cole is English. ...pulling from memory. I need to find the source to verify that last part.

Marcus told me that their products worked without factory OC. What then if the client wanted to OC, at your own risk.

From my communication with KnC, the box will come to them from the factory at stock. KnC will apply some overclocking themselves, then ship the miner to the customer. I was told the customer would be able to OC, but obviously would void the warranty in doing so.

[Bitcoinorama]

Look, this is silly, because it's obviously a mistake and it's a very simple answer because most of you above have followed this the entire way through and when Jupiter was announced it was 250gh/s and the 30% upgrade info was already being touted, just not applied.

Then they introduced Saturn and revised figures, at which I point I know for a fact I asked and gave the answer in the main thread. So if it's worded otherwise it is an accident, as well most know, and that they are already breaking ground with what they are claiming to offer, which is unmatched.

I suppose if you don't want to dig or follow the communications published in the forum in their main thread, and prefer to argue semantics, you're always welcome to a refund...

EDIT: just looked at the product page again, irrespective of the additional 30% gain to the algorithm, the only thing that matters is the promised minimum 350gh/s per device.

[brontosaurus]

Forget about all this speculative stuff. If I go to buy a car with a 100 BHP engine, that's what I get. Same with a 2TB hard drive, a 32" waist pair of jeans or a 400g tin of beans. If you don't get what you pay for you would be angry, and rightly so.

In real world engineering, we work in specifications, not speculation. Would you buy a car that 'might' give you 100 BHP? Of course not, you want the figures pinned down tightly as to what you are actually going to get. Yes, you could possibly tune the engine or use Nitrous oxide to 'overclock' it, but both measures might destroy the engine by overstressing it beyond it's design limits.

So demand a proper specification for any miner device - not speculation as to what it 'might' do. If the designers can't give you a proper answer to that question, then you should be very wary. Modern software tools from companies like Cadence can predict very precisely how silicon will behave once manufactured in a particular foundry's process, so as long as the design has been done properly there are NO excuses for not pinning down the specification.

BFL underestimated the power consumption of their asic by a factor of six. This should not have happened, but I suspect they based their figure on an asic process but effectively used an FPGA transfer, using up a lot more gates in the process. On the limited information I have on their device and die size, in a 65nm process they should have a single chip capable of around 11 - 13GH, not 5 -6.

[Bitcoinorama]

Precisely Minimum Specification promised is 175 gh/s for the Saturn.

Minimum specification for the Jupiter is 350 gh/s.

First line on both respective pages. Only thing that matters.

The 'water-cooled' set-up bit should come off though. That is feasible and they will offer an option for those at home if they want that at a later date, but data centres have in no uncertain terms said absolutely not to housing any liquid cooling.

[DPoS]

They really should go through these threads and update their FAQ on their site if they can find time - then these threads can calm down a bit

[Bitcoinorama]

They really should go through these threads and update their FAQ on their site if they can find time - then these threads can calm down a bit

I know they're updating it all next week. There is a complete revision on all specs (positive revision), with a more accurate power estimate, which is why they said to hang back on picking up a power brick in that PSU thread yesterday...

https://bitcointalk.org/index.php?topic=237745.msg2520109#msg2520109

[jelin1984]

Any news about knc miner ASICs chips? Or anything?

[Loredo]

Forget about all this speculative stuff.
...
In real world engineering, we work in specifications, not speculation.

+1