Topic: Swedish ASIC miner company kncminer.com - page 1935. (Read 3049501 times)

its ok, all the knc supporters tell me that its not an issue. Since they dont care about power i guess, which leads me to the conclusion that they have unlimited power capacity at their mining location. Oh ya they will also say that the power usage predictions are the worst case scenario so it can be better than bitfury when released.

Ummmm. KnC is doing a 28nm process and getting only a third of the GH/s per wafer that bitfury is getting at 55nm?

Be careful dude, I made that question for weeks and I was lynched by the wishful thinkers and wet dreamers

ok, makes sense to me, thanks.

It's the SKU code.

http://en.wikipedia.org/wiki/Stock_keeping_unit

Another detail for a better table. As far as I know, 130nm(110nm) are still manufactured based on 200mm wafers. 65nm(55nm) nodes were the first built with 300mm.

Preamble: I am relatively certain that KNC Miner is not a scam. (about 75% certainty) I have serious doubts that they will meet their targets/estimates in terms of hashrate and delivery dates. (less that 50% certainty) but I am considering taking a risk on the new Mercury device... (pretty affordable entry level device, limits my risk to acceptable levels.)

Question: What do the little grey numbers below the various products mean? You can see them from the "Miners" tab on this page -> https://www.kncminer.com/categories/miners

Currently there is a 12 below the Mercury, a 34 below the Saturn and a 250 below Jupiter. Does this number mean anything? I assume it has *some* kind of meaning, or else why is it there?

If I were to assume I would guess that this number represents the number of orders, but I would appreciate some clarification on this before I throw 25+ bitcoin on this gamble.

Any educated guesses are welcome, but clarification from a KnC official would be preferred.

-Thank you.

Thank you HyperMega. This thread really needed some objective and useful content. 

Sorry, but this KnC die areas ist most likely not correct. The table should be corrected.

There is an easy rule of thumb to estimate the silicon area of a pipelined SHA256 core in 28nm. From every technology node (half node) to the next, one will get approximately 2x more logic on the same area.
From 65nm(55nm) -> 45nm(40nm) -> 32nm(28nm) you will get 4x more logic on the same area. Meaning you will need 1/4 of area for the same logic.
BFL die is about 50 mm2 with 16 cores. Removing some overhead for pad frame and supporting logic a single core in 65nm should be about 3mm2. Scaling this down to 28nm would result in 0.75mm2 per core.

Based on my experience an automatic place&route with an high density standard cell library gives you in many cases better results in terms of area/power/timing than this "old school" hand layout done by BFL (sorry Josh ), especially for advanced nodes implementations. That's why I assume 0.6mm2 per KnC core.
For 192 cores (engine IPs) this results in about 115 mm² overall die area, which is still huge for 28nm (equivalent to 460 mm2 in 65nm).
Ok, at the end there are still some mm2 to add for support logic, but for sure not more than 5mm2. So in sum not more than 120 mm2. This is what they have to reach, otherwise they would not take use of the full advantage of the 28nm technology.

The complete calculation is based on the assumption that an KnC engine IP is more or less the same as a widely known pipelined SHA256 hash core. If the KnC IP is more efficient, the die area could even be less than 120 mm2.

Die size is less than 336mm2.
I think 18x18mm

"Damned if you do, damned if you don't!" is said very often here.
But "Damned if you do it not right!" would fit much better in this case.

Why did KnC publish these technical slides without any comments? In this way they say almost nothing about the feasibility of their concept and the overall project status. Just a nice slide show to let customers with no or little technical background feel good. This automatically causes a lot of wild speculation. So why not add enough information to make the slides self-explanatory, e.g.:

Slide 2: ASIC schematic
192 (48 x 4) engine IP cores (pipelined SHA256 hash cores, one hash per clock cycle)
-> running at minimum [email protected] to realize 100 GH/s per die (could be higher in case of defect cores)

Slide 3: ASIC Toplevel Floorplan View
-> ASIC requires hierarchical layout flow to handle complexity (millions of standard cells)
-> complex engine IP (multi-million gates) "hardened" first; seen as multiple instantiated cells at ASIC toplevel
-> 0.6 mm2 per engine IP results in about 115 mm2 die size
(BTW: PLL arrow points to wrong location, should be probably rectangle with red halo at the right side)

Slide 4: Engine IP floorplan
-> engine IP hardened 2x with different shapes for area optimization at toplevel
-> bigger notch of 2nd engine IP macro used as placement area for PLL at toplevel (2nd macro placed 2 times at the right side of each quad)
-> "magic" pattern caused by timing driven placement of standard cells related to pipeline stages; just a funny visual effect

Slide 7: Thermal simulation
-> Assumption: 200W power consumption (based on worst case power analysis @ 1.0V & 30 % over clocked)
-> Simulation result: die junction temperature does not exceed 125C

Disclaimer: These are no official KnC information. Just a proposal and educated guesses of an ASIC design engineer how to publish such stuff in a more professional way.
 
Anyway, the most important question regarding the project schedule and the feasibility of the announced start date for delivery is:
When was the tape-out of the ASIC executed? If not executed yet, when is the tape-out planned for?

If you think about it. Payment was made from a US company (BitPay) for a delivery in EU . No VAT

These days you aint sure about nothing... 

No problem.  If chips run too hot company changes name to KFCminer...

       

LULZ

This is my KNC identity. I have others.

And who's sock puppet are you again? Seems almost all your posts since you registered last month are about KNCminer.

Darn.  Apparently even eve is a dude. 

Looks like we got them guys. They put a lot of effort into looking like a real company. Partnering with a proven manufacturer, all the lies and deceit. We were all fooled. It has all been a scam, they don't even know how to fake temperature simulations right. Either the chip is on fire, or it must be run only in the winter cuz the scale must be farenheight.

Orsoc was never working on any asics, it was all a scam, sure they would be throwing away their reputation for what looks like peanuts, but sometimes it's more about the scam than the loot.

The mars prototype just shows you how much of a long game they have been playing here.

All that effort, just to be caught on such a small thing. It was almost the perfect crime, thank you dogie for opening our eyes.

tl;dr - sarcasm

I paid with Bitpay and the VAT was not substracted, I had to pay $ 2,504.25, for delivery inside the EU.