Topic: KnCMiner declares bankruptcy due to increased competition, upcoming halving - page 2. (Read 4263 times)

the pci-e spec is for computer cards only. so to be within spec for computers its 75/6 pin 150/8pin

the connector itself is rated 150/6 pin 300/8 pin as other have said.

Good discussion of this issue here http://www.tomshardware.com/forum/274631-28-power-spec-power-plug#2674391

The high current pin currently in use with 16 ga wire allows ~396 watts on the 8 pin PCI-E power connector.

I've had a Seasonic PCI-E power socket melt on me and also a KNC Titan power socket, at different times, though both times I was using the KNC supplied splitter calbe. It just depends on where high resistance due to corrosion or oxidation occurs in the system. I now periodically remove and inspect all my Titan power connectors for signs of problems and put them through a couple of insert and remove cycles to wipe the pins and sockets clean.

I RMA'd a cube to KNC with this problem and had a turnaround of a few weeks. I paid shipping there, they paid shipping back.
They always were tight with their money ...

I rarely post here anymore but I will say that Sidehack is the go to guy for mining power supply solutions.  I have half a dozen of his server power supply adapters, they work FLAWLESSLY and never an issue with overheating or malfunction. His work is beyond reproach as is his reputation.

More info on this, translated from Swedish:
http://digital.di.se/artikel/knc-agaren-vi-hade-inte-koll-pa-produkten


So I guess kfc fucked it up technically, as well as all the other bad decisions they made.

75W on 6-pin, 150W on 8-pin. Course 6-pin only requires 2 of the 12V wires to be there; 8-pin requires all 3. The typical PCIe crimp pin is rated for 9A, which on a 12V line and 10% safety margin puts you at a shade under 300W. Course at 9A, with a 2-foot cable of 18AWG you're losing 1V (and about 8% of your power) in the wire, making a lot of heat; with 16AWG you're still dropping about 0.65V (5% of power) and a fair bit of heat. So pushing a connector to 11A per wire/pin is definitely asking to fail.

Which drew more power per cube, Titan or Neptune? I recall the Neptune was spec'd for 2100W across 5 cubes and controller, but a hosted Titan I had here ran 5 cubes off a 1600W ATX.

I've also always assumed KFC's 16nm claim was just a bluff. I wonder if some of their technical difficulties could have been solved by running a grid of small chips like almost every successful miner has used, instead of one giant BGA device like almost every failed miner has used.

Yeah. My view is KNC orig design was spec'd out as 250mh miner in original announcement. Then because most engineers overbuild when they were 3-4 months late they oc'd the Titans to 300mh because they had this slack to do so. But being greedy dorks they are and imho because they stick thought these Titans would be under 90day warranty only.  (Not so the forgot to change terms of 12months on web page from the previous btc Neptune miners. Thus screwed themselves.) they then oc'd it again to 350mh and thus therma issues and need for split y psu adapters.
The need for 350mh was with all the rma's requested after launch they had messed up promising 300mh. My unit at 300 mh setting qualified for rmA for 2 cubes. With the 3rd firmware update to allow up to 350 mh speed settings. I got 306mh and thus refused Rma o 2 dead dies on 2 cubes. But as a result that psu plug is byond spec and melts w/o the psu y adapter. Hell sometimes with. I just think it funny that it was a great plan on 90 day warranty. But not so much when they screwed up an had to honor rma's for a year. Oops

Doesn't the PCI-E 'spec' call for a 75W "max" on 6 pin?

I posted about this in the main kfc thread, their investors claim big technical problems
https://bitcointalksearch.org/topic/m.14997169

Either it was a massive bluff about 16nm or they have 16nm, but they are massively failing/underperforming. Either way, fuck 'em. They deserve some karmic payback

Very good question. Is there any proof that they really have 16nm-chips? Their share of the hashrate is/was small...

Anybody with inside knowledge?

Just curious, what went wrong with the 0.06 J/GH 16nm Solar ASIC?
They claimed to have it since when? Beginning of 2015?

How can a company with this piece of “alien” hardware loose the mining competition?

Even if Bitfury delivers its 16nm ASIC in about 2 weeks , the Solar ASIC would still be more efficient!
And even including Swedish energy taxes they should have <$0.06/kWh power costs in northern Sweden!

Was this Solar ASIC just FUD?
Was KNC not able to create a working miner based on it?
Or is the ASIC too expensive caused by low 16nm yield?

Hopefully they do mess with the wrong players - eye for and eye

highly  Possible and they know it then come out on top smart move and riskily but smart but at some point they will wtf up and mess with the wrong players if this happens...

You don't have to be Swedish or know anything about Sweden to find nationalist bigotry risible and objectionable.  As if there were actually an Swedish or American "business mentality" generalizable over millions of citizens (or even a couple of notoriously freethinking Bitcoiners!  ).

Regardless of the backpedaling over those secondary matters and process/meta concerns, I believe what you're saying about potential improvements to the chassis.

But it's not reasonable to expect KNC to have made perfect the enemy of good enough, nor is it fair to assume they would have necessarily made all the best/correct calls on the trade-offs involved in the rushed design process.

I'm sure you were willing to wait with heroic patience while KNC got everything right.  However, the rest of us (individual and group/institutional buyers) were counting down the seconds until the damn things shipped!  Thus we demonstrate the internal logic of the task's thankless nature, and why eventually someone always winds up getting called "human garbage."   

Some of us even invested in mining companies, and so had in terms of calculated risk 3rd order exposure (and in terms of gambling, a risky bet on a risky bet on a risky bet).   

$585 at Huobi atm.

Good riddance KNC.

I ignored a blanket condemnation of Sweden, made by a Swede, in another thread, because it's not something I know about. If someone wanted to trash the American business mentality, I might speak up - unless I thought his assertion was valid.

I defended a comment about "human garbage" in another thread because, as a person who doesn't like to be stolen from, and as a businessman who works to provide for my customers whether I win or lose (and I have taken my share of losses to keep promises I made, losses caused by others failing to keep business promises to me), I don't think it's entirely fair to blame the victim of theft for allowing himself to be robbed and give a pat on the back to the robber as "just doing his job".

KNC definitely was winning for a while. Their silicon design guys have always done a good job of keeping ahead of the curve on efficiency. But you can't convince me that building a case from 18AWG steel instead of 24AWG steel would have lost them the race, or that putting a second PCIe jack on the board (which, really, should have been there from the start) would have lost them the race. I don't expect carrier-grade, but I do expect machines that survive shipping intact and are designed to not catch fire. There's a reason my weakest PSU board still used 9oz copper where some other guys' strongest boards use 3oz.

As someone who has been building power supply equipment for 2.5 years, hosting miners for 1.5 years, designing and building miners for 1 year, and working my ass off to fulfill business promises for five years, I like to think I know a thing or two about building things right and how to not screw customers.

I also don't think you know me well enough to judge what I would and would not be complaining about. And yes, I know a thing or two about "best efforts" and "thankless/endless tasks" especially in the bitcoin economy. And I do not backpedal.

KNC was winning the 3rd gen ASIC competition for a while, thanks to their rapid time to market.  That successful execution didn't happen because they were scamming or "mailing it in."  

I can easily bore you by reminiscing about the details of the race to 28nm, if that's what you want.

If KNC took their sweet time and your sweet money to painstakingly design/test/deliver bespoke "carrier grade" equipment you'd be complaining about that instead, as it would have made the miners much less profitable.

KNC obviously put in their best efforts given the time constraints and trade-offs involved, which ironically was a mistake, because dealing with retail gamblers is (as you continue to demonstrate) a thankless/endless task.

You've gone from defending comments about "human garbage" and ignoring blanket condemnations of Sweden to niggling over build quality.

If you backpedal any faster, you might start to polarize the vacuum itself, like a very small magnetar...  

How did those Y connectors prevent melting when it was the same amount of current going thru that 1 PCIe connector on the Neptune?

I'm pretty sure you've never had any Neptune or Titan.

I had 6 x Meptune set. 2 melted PCI-E socket before I got Y spliters. After splitters none  burnt  PCI-E connector.

I have today over 30 Titan cubes. All with splitter cables. No burnt  PCI-E connectors an cube but some burnt   PCI-E connector an original KNC Y cables.  Not  originals working very well.

Since the Cube cooling fan cools the PCI-E connector , does not have any connection with PCIE spec  300W max.

The problem is not the cable. I never said it was the cable. Nobody said it was the cable. The problem is that the miner itself is drawing up to 400W through a single PCIe jack. People sometimes have trouble with the SP20 and Avalon6 taking in 500W through two jacks and still roasting connectors. The KNC Neptune pulling 400W through a single connector and causing fires is a well-documented problem, not something someone made up on a whim to make the manufacturer look bad. Just do a quick search for "Neptune fire", you'll see claims going back to 2014. That's the kind of thing even an intern engineer should have caught long before it came to production.

I've got some pictures of my own, of some Titan cubes that required rebuilding (for at least the second time) after shipment because the entire housing was so flimsy that heatsink mounting screws were ripping the threads out and coolers were flopping around loose inside the case breaking stuff. That's also the kind of thing even an intern engineer should have caught long before it ever went out the door.

Aside from any complaints about how the management may have mistreated customers, the design and build quality of their later products is some evidence that they were "mailing it in". Those amateur engineering blunders are the only things I have hard evidence for, but it's enough to convince me they stopped caring about what got sent out a long time ago.