Topic: Swedish ASIC miner company kncminer.com - page 1120. (Read 3050068 times)

I can't comment for KnC. However from what I recall from one KnC news post. The original designed boards had 8 vrms and in the article they had a comment that it would play a key role on 2nd gen design chips.

However they may have had a set amount of those boards made. And before the tested chips arrived may have commissioned the new asic boards and control units with less components. Perhaps a cost saving measure.

remember this??   isn't there a 40 amp spec limit on them?  correct??   #3 below has 8 VRMs and looks fine,  the others pretty baked.  So stop saying they werent needed..   perhaps not to power the chip but to regulate the power, yes

not sure why everyone is missing my point.  They did have a working miner and shipped X number of miners with 8 VRMs.   So before they go cutting costs they should just test and make sure they aren't frying their product right?

What is everyone's point on 'time sensitive tech' etc. when the reason was they wanted to slash costs before they tested a stripped board?

just own up to that decision KNC

Cool, glad you're happy. If Avalon had delivered it may have cost you more.

They were needed originally, but they were an necessary margin as they weren't needed to power the device to spec? I'm not pushing anything under the rug.

Dude i'm not a software engineer, I have not designed the hardware. I personally haven't hashed a machine fully before .95. You can follow the link in my sig and see my hashing as i've played with rigs.

I have seen so many revisions of software and hardware, including a hell of a lot that won't be seeing the light of day in public. So I do not have the intimate knowledge of each particular revision like yourself who has had longer to play on each, or the software engineers that have written and redesigned the firmware, or those making the hardware decisions. You are expecting me to be an expert on others decisions. The hardware i'm playing with next to me doesn't have 8 vrms. I personally haven't played with those. I've remove, tested, and replaced some when working through some RMAs, sure.

Honestly most of this has been a insomniac, food deprived, stressful emotional blur moving at light speed for me. Right now I'm staring at a screen exhausted at almost 10pm on a Friday night having Q's I do not know the technical nitty gritty to fired at me. I was in meetings at 9am. I'm taking the time to answer you out of free will, not so I can spin a web of deceit. That's not some flakey BS, but no one is using 8 VRMS currently to obtain spec with current firmware. They are not necessary.

The real problem will be all those that increase their mining operation without selling a single unit: Bitfury, 200 THs or whatever it is right now, Avalon team said today they are going to be able to deliver 2 PH of chips monthly, BitMine.ch has a 4 PH private mining operation to build plus whatever they will ship, BlackArrow has 2TH/s units going online at the end of february next year, HashFast will deliver, sooner or later, and so on.

So, the fact that difficulty is lower this or next difficulty jump does not mean we're reaching plateau.

spiccioli

I agree.. I lost one die from it all, but mostly they survived.  But you know that was not my point in bringing it up.  It is that certain people are acting like there was no problem removing them (no effect at all as he says) and then wondering why everyone was blowing fans on over spec VRMs

The problem that persists with any time sensitive tech is should we make it perfect or ship it to the people and update it as we go.

The community is well aware what happens when you try to make it perfect, with some vendors delivering 14 months late.

Even Bitfury systems were not Perfect. One good thing Punin / Dave did was send extra GH/s to make up for it.

I know what it means to pull 9 day work weeks. 16 hour + work days wears down common sense, my only gripe was that I had to step in for KnC's customer support. I could only guess how many mixups they had on day 1 orders.

My concern stemmed from the lack of recourse a colleague who had lost 42 BTC due to an admitted queue mix up on the part of KnC and had his orders pushed back well beyond day 2 orders.

Everyone expects compensation. The reality is that those who had day 1 orders. They expected to receive their units mid September 2013. So I can understand those who are not thrilled of the delays. Perhaps the customers expectations were unrealistic. However KnC have made it up for some.

tl;dr sure Sam and his team had issues but in the end they delivered as well as any one would hope to expect.

This makes me feel so much better, late deliveries and constant failures have cost me about 26btc. KNC's late and tardy delivery have cost buyers close to 20mill in lost USD, but at least it was in record time!

Let me quote some things for you:

Our complete services for Bitcoin mining make it easy and cost efficient to mine Bitcoins. Smart solutions for everyone who wants to mine for Bitcoins.

>>First sentence when you google KNC

Das riesige Gerät 576Gh/s-KnCMiner Jupiter ist der mächtigste Bitcoin-Miner, der jemals gebaut wurde. Es ist eine Maschine, die für jedermann zum Kauf zur Verfügung steht.
Die Tatsache, dass die Öffentlichkeit die Plug-and-Play-Geräte erwerben und betreiben kann,...

>> official German press release, feel free to google translate.
 That took me 2 minutes, there are way more news out there, even audio files of them statiing everyone can buy their miner. Plug and Play is also stated here.

KNC has voided their own terms of servie with their official statements, and now you want to come up with it's not a cosnumer device and not consumer friendly?

Let me translate the above for you:   "It's a machine anyone can buy. The fact that the public can buy and use those plug and play devices..."

Do you have anything to add to your statement?

Well good thing GE makes their shit good and the VRMs didn't catch on fire and even if there was a real danger there it seems like nothing went so horribly wrong as you would expect. 4 VRM boards lived enough until post 0.95 firmware.

It's only good for Bitcoin to have more and more miners

how do you completely forget about the proof that 4 VRM boards were COOKING well above 40 amps with pre 95 firmware while the boards with 8 VRMs were not running amps over spec?

you still act like they were just totally dormant 4 VRMs removed when the design with 8 VRMs handled the power in spec with pre 95 firmware and the 4 VRM firmware didn't

so until they had it under spec with firmware 95, they WERE needed

that is what you keep trying to push under the rug in all your responses to this

Act like what never happened?  I'm not playing dumb, I'm flat out exhausted, ref. above, really.

The 8 VRMS were a margin in the beginning. You obviously reduce margins as and when you are confident they are unnecessary. At that point in time they may well be usable in fixing products if further problems arise and excessive replacements would have been needed.

As I said the best analogy is it's about having a family of components that have never met in person and will likely disagree behave with one another. You try to over compensate to appease. Those with 8 VRMS have 4 that are turned off. It's pointless to continue shipping components on a board that aren't being used. In future, boards may well have 8, the physical space is there, and options still exist, but the compatibility has to also be present, and refining that compatibility is what consumer device manufacturers have the luxury of 12-18 months for, and we didn't. Bitcoin does not grant you that currently. There is no room for failure, or forgiveness. Not in the race that's just taken place. You cannot afford a major design mistake, you're finished.

That's another six months if you have to rewrite RTL from scratch, 12 weeks if you need to reorder new parts.

It wasn't fuck them till we fix it, it was more we've fucking promised this, and we will damn well fix it. For the most part it's always been about firmware, but no one wanted to believe that. I'll be honest I had my fingers and toes crossed there, but I didn't design the hardware, and they overcompensated for the reason hardware could handle it, firmware could fix it.

I will never forget how white Andreas turned at one point in the manufacturing facility on the day of the delivery for a schoolboy error. F-ing classic. At the time it was full on brown trouser, no laughing matter, but the colour just drained in seconds for something so simple, so looking back it's hilarious.

Sleep, wow how I miss that, in all seriousness I've literally been delirious and emotionally fried from the lack of it. Honestly.

I let you off for a while, but you still haven't come clean and act like it never happened.

BUT...  you say it 'wasn't refined' but then WHY did they pull out 4 VRMs before they either bothered to test them, or knew and just shipped anyway??

Why not wait till they had proper firmware (.95) to at least handle the power issues over spec before stripping the boards?

I agree to let a lot of the other nonsense slide (what happened to next shipping? no ship on sat??? what??) but to act like a big 'fuck them till we fix it' decision wasn't made is LYING

so just own up to it

Why would the Nov batch be smaller than the Sept/Oct batch?  If KNC knows they can ship X units in ~30 days the first month with the most problems, the most unexpected delays, and the most stress why would they anticipate only selling a fraction of that the following month after having 15 days to mass produce boards, catch up on some sleep, improve their shipping process?

Would you like them to sell at that volume?! I mean those factories exist sure, but they want you to ship thousands upon thousands, if you're going to use them. Logic dictates that's why they build those premises...

No offense, but shipping a couple hundred units a day is amateur. Professional factories can ship thousands upon thousands.

No, this is not a consumer device, neither is it consumer friendly, they are however incredibly busy constantly trying to tweak it and improve it, which is obvious as there is a reason atypical 28nm devices take 12-18 months post fab to refine. This is the fastest 28nm device ever created. The surrounding components are like a family that doesn't get on with eachother, but have to meet a new family member for the first time and keep up appearances. They've picked the strongest personalities capable of managing this and placed them in a room together, but really and truly they've never met the ASIC before. Will you wait 12-18 months for a device to be ready? For a company's customer service to be prepped on it inside out?

No, this is Bitcoin you all want the tech now.

If you want a consumer device like an Apple iPhone, then understand Apple have both the iPhone 6 and likely the IPhone 7 in hand are tweaking it to perfection for release. KnC haven't had the luxury of that time. you've had the damn thing for a couple of weeks less than they have, and we're all finding out what works, and what doesn't as an iterative process. The company is actually full of the most solid work ethic i've seen it's half 8 on a Friday evening and the software engineers, myself and most others are still here. No one has relaxed, sure the pressure and burden to some extent is off, but no one has the attitude of "Got your money..fuck off there's a good chap". You're just making crap up you want to believe. Really.

As I said months ago Marcus would have played with that thing and delayed it's release to perfection. Sam and Andreas understood for you have to have that device hashing in your hands asap. You got it, unrefined, but as refined as they have had it themselves, shipment began that same week. Issues are dealt with as they arise, and the majority are over the moon.