Topic: Swedish ASIC miner company kncminer.com - page 164. (Read 3049514 times)

Executive summary: KNC are a bunch of blighters

Where did you get this quote Avenger?

The clown acts publicly as if there is only a single customer with a single legal issue involved.

There are some 28nm folks that got shafted by their ineptitude at execution.
Someone @ kfc determined that shipping miners with fans held in place with twist ties, suitable for bread bags not international shipping was a good idea.

All 20nm CA batch customers know by now sammy &Co. put sand in the vaseline.

IMneverHO (<<--notice the huge disclaimer)
I think the whole current situation boils down to 2 very poor decisions by knc management.
First was bonus miners to stem the bleeding from Nep refunds.
This decision committed them to needing 33% more ASICs for 'remaining' CA customers.
I think their mouth wrote a check their ass can't cash.

Second was 5 ASIC per Nep instead of 4.
This decision committed them to needing an ADDITIONAL 25% more ASICs

They could have redone the PCB to include 12 DC/DC instead of 8.
But then they would 'loose face' with their power claims.

The difference is
4 ASIC and 48 DC/DC.
vs
5  ASIC and 40 DC/DC.

If ASIC were as common as DC/DC converters the decision would just be a matter of cost.
If an ASIC cost less than 8 DC/DC converters, it favors the 5 ASIC design.
I can buy 8 DC/DC converters for under USD$ 200 without a volume discount BTW.

The problem with this whole idea is that ASIC are NOT common.
As a matter of fact they are almost as rare as unobtainium.
Add the fact that their value to the customer RAPIDLY decreases over time.

I don't often 'should' on folks but...
knc SHOULD have done the original PCB artwork for 12 DC/DC converters.
This would cost no more per unpopulated PCB.
knc SHOULD have placed 2 power connectors on each PCB.
(cost more for 6 additional holes drilled and via's installed per PCB. (a few cents per)
Their original Oct 28nm PCBs WERE laid out for 2 PCIe power connectors for fucks sake,
it's not like the concept is foreign to them.

knc's whole relationship/life with DC/DC converter implimentation should accompany them to their approaching grave.
A large percentage of their software changes have been DC/DC related.
They have referred to the issues as 'ghosts' demonstrating their understanding of the root cause.
They continue to have issues with sleeping(can be ressurected) DC/DC converters.
Every block we roll the dice on dropping a die from a sleeper.
Then we loose 100% for a minute or so as the whole thing recovers.

It's been 6 months since 20nm launch.
It's been 6 months since knc determined actual yield.
They know WTF their obligation to customers is/was and have come up short.
The fab HAS tapeout already, couple months to make more wafers.
They did it 3 times now 28nmBTC, 20nmBTC, and whatever the Tit is.


My final bitch 'this post' is clocks.
KnC set the bar for customer expectations with their 28nm product.
They did not go out of their way to obfuscate clock settings.
knc DID NOT DELIVER minimum expectations for this customer re 20nm.
With the 20nm product they INTENTIONALLY artifically limit customer clock settings.

We KNOW the 20nm code is intended for optimization/maximization of clocks on the fly.
I'm left thinking they reserve this extra 'x' % improvement to skim off of in house product.
There are ways to get more out of each customers ASIC.
They have been in place the whole time.
The window for a few % making a difference closed 1-2 diffs after launch.
The OC rule for BTC is early on matters, later not so much.

Why would they do this?
See DC/DC quips above^^^
The fuckin things are on the verge of flames now with the single point of entry for power.
If their earlier claimes of safety are true...
The "Y" cables should come with some extra clocks.

knc brought a win/loose game to a win/win situation.
It's no suprise that the customers that end up loosing are now playing a loose/loose game.

Their cash flow situation 'appears' to be extreeme.
Why else would they bleed their customers refund cash for 6 months?
(knc appologists feel free to correct with accurate count of MONTHS the last guy waited)
I doubt they have the cash to even SHIP the bonus miners.
Add the fact that many in 'destination tax situations' will refuse shipment as they will be worth less that the tax owed.


I don't care if the cause is ineptitude or evil, the result is the same for the customer.
They might claim there is only one guy with a single issue.

I counterclaim they have not produced a single satisfied 20nm HW customer.
Not one!


YMMV

Welcome to 14 months ago when they shipped the die 0 crocks

kfc have always and still do the bare minimum to fulfil their obligations and then endlessly promise firmware updates that will cure everything. Nothing has changed in 14 months, only a few more people have seen through them.

For giggles:
https://translate.google.co.uk/translate?sl=auto&tl=en&js=y&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fcomputersweden.idg.se%2F2.2683%2F1.602340%2Foj-nu-blev-det-dalig-stamning&edit-text=

Their new Orama is too busy to answer any questions, but eventually says (via google translate):
"We only know that a customer's claim for compensation and threatened legal action. KNC Miners attitude is that this customer demands and threats are unfounded. We handle general requirements of our customers according to established procedures at the pace they alleged and act now as before in accordance with the conditions we have agreed with our customers. We comment in general, not rumors, competitors or alleged / current / potential legal process, "writes KNC Miners marketing and PR director in the mail.

So stop complaining everyone - you agreed with kfc they could ship you faulty/barely safe/functioning hardware  

Thats what I was thinking as well. My cubes are properly cooled  and can be pushed some more. Just want to try squeeze as much out of these before they become paper weight

that may be good to know I for example have 1 cube that runs COOLER then all my other cubes
dc/dc wise etc at 325 stable and has the fastest results of all my dies/cubes

I figure if there was a setting for 350 it probably could pull it off (the other dies/cubes/etc er...not so much)

Anyone over clocked the Titan beyond the 325 MHz ?

Thanks Tolip, i'll give it a whirl. I would not attempt a RMA with KNC except in the most dire of circumstances, and maybe not even then. Even more than the pleasure they seem to get from swinging the Viking ban hammer wildly on their forums, they seem to love the power of making customers suffer through insanely long RMA wait periods.

Just got the Y cables for the titan, 6 weeks after the titan's delivery date (I didn't accept the delivery due to the lawsuit)..  since I don't need them, how much do they worth ?  

Thats cruel and cold them shipping < 100% working miners,
of the few they manage to ship...

! you might have a look inside the cube with a bright light and a magnifier.
without opening the cube it is possible to see if the heatsink on the DC/DC is the issue.
# 6 & 7 are on one end of the "E" plate.
(assuming they used the same setup as for the nepturd)

I posted 2 methods to remedy this without any spare parts.
https://bitcointalksearch.org/topic/m.7727618

Might be worth looking, instead of waiting x weeks for RMA.


YMMV

 That's much better!  A picture is worth 1000 words in any language.

Jelin, never mind cranky old homer. AWOL means something isn't where it should be when it is supposed to be, in other words 'absent without leave'.
If you look at the advanced page for my Titan, there is no entry for one of the DC/DC converters.

See how DC/DC #6 is missing for ASIC 3 in the image below ? It is AWOL.

http://imgur.com/IoJdvKx

 The guy took a military abbreviation turned it into an idiomatic expression that doesn't really even make good sense and you think that googling AWOL is going to help someone with ESL?

http://lmgtfy.com/?q=AWOL

What mean AWOL ??

In my case, one of the DC/DC converters is AWOL on the advanced page and the associated die won't run at full speed. I expect to find a bad solder joint or a cut trace on the PCB or something.

POS quality control from KNC. I've read stories on their forum that they are offering to replace some customer's bad cubes with multiple partially working replacements …

POS company. POS products.

I try but nothing happen

Thats good news. I've been thinking of doing the same. I'll take photos.

Really that is very nice is it true
Any photos?

You mean that the fan not be good at the chip?

Just opened one of my titan with 2 die down.
The termal past was dry and not applied everywhere.
Removed it.
put artic 5.
break the head of a shitty screw.
find a replacement screw.
4 die working

Whats knc's wallet address?