Author

Topic: Swedish ASIC miner company kncminer.com - page 356. (Read 3050076 times)

copper member
Activity: 2898
Merit: 1465
Clueless!
This shit is crazy.... It really looks like they are mining with our machines RIGHT NOW.
I can think of no other way that hashpower could possibly magically appear
https://blockchain.info/pools?timespan=48hrs


Looks like a BFL replay...   https://bitcointalksearch.org/topic/i-no-longer-work-for-butterfly-labs-bfl-659828
The drama of it all....  Holy Swashbucklers.


it is simple ...if you hash while you wait we get your 2nd machine the was gonna be compensation (but we don't wait 3 months for it)
then when you get your machine in 2 weeks the above hash goes back to knc farm

if you don't hash while you wait....we (maybe?) hash while we get a machine out but most likely just get you your machine in line

if you have refunded...we never bothered to pull the equip out of the que.we stall on getting refund to you and then take your
equip (prob a head of everyone else) and put it in the PH farm

either way...all refunded machines that were paid for ...are still being delivered to KNC imho just go to the farm

all GH hosting set aside for customers that did not take it go back to farm or did take it go to farm after we stick a tracking number on the  box and out the door to you

they likely have 2x or more machines made 1x is customer orders (all the above) other lot is KNC farm orders ..to be plug'd played
into the old jupiter racks (i bet boards are just snap/screw in and viola ..replaced the other half is what ships to us (specific to
the considerations i stated above that PH farm has priority)

the cool thing about the above is the 'slow' refund rate to customers but the real cash used to get chips for equp for the machines to
go to ph farm asap w/o interuption...even if customers cancel order....no hicups in the pipeline to the PH farm

as far as i think it is all streamlined to get PH farm equip/hast asap..and with likely over ordering with 'not so fast" pre order $$$ it is likely this is working well
and anyone who want so GH farm on the cloud hosting their 'compensation' machine likely already has been figured into the above equation and it to
jumps to the head of the line

with KNC's jump in share of network at the VERY LEAST they ordered 3x the macines (modules) for themselves vs what they put out for customers imho
the stuff just did not pop out of thin air imho

Searing


full member
Activity: 203
Merit: 100
Here is a datasheet for a typical remote diode temperature management/fan control IC.

http://www.ti.com/lit/ds/symlink/lm96063.pdf

Note: "The LM96063 is remote diode temperature sensors with integrated fan control that includes remote diode sensing. The LM96063 accurately measures: (1) its own temperature and (2) the temperature of a diode- connected transistor, such as a 2N3904, or a thermal diode commonly found on Computer Processors, Graphics Processor Units (GPU) and other ASIC's."

These type of devices typically integrate with an on die thermal diode.

On the issue of BRM46x thermal design constraints, the spec says 125 C. Not 100 C, not 105 C. Looking at the data sheet, it looks like the device temperature is reported via the PMBus and may also be subject to configuration via a configuration file:

"This product is designed with a digital control circuit. The control circuit uses a configuration file which determines the functionality and performance of the product. The Electrical Specification table shows parameter values of functionality and performance with the default configuration file, unless otherwise specified. The default configuration file is designed to fit most application needs with focus on high efficiency. If different characteristics are required it is possible to change the configuration file to optimize certain performance characteristics. Note that current sharing operation requires changed configuration file. See application notes AN307 for further information."

Is anyone having problems with Neptune's shutting down due to thermal issues or is this yet more FUD being spread by haters and envious ne'er do wells towards KNC ?


Thanks for the information on the LM96063 and pointing out facts throughout this thread. I work with equipment where I can measure temperatures 100s of yards away so I don't understand why anyone would thing a small IC on the side of a board is not capable of reading temperatures directly inside an ASIC several inches away. The more I investigate and sift through the postings of people who don't really know the more comfortable I am feeling with the Neptune.

I have no doubt that a better heat sinking arrangement for the VRMs is in the works or could be made. At the least it's an easy thing to modify and I'll get my 12 year old (because of smaller fingers and  better eyes) to take care of it for me. Based on the specs I would like to see temperatures maintained below 100 and preferably 80 to 90 degrees C on the hottest VRM.

Yes: I'm already working on this.  Many things can be improved.
member
Activity: 84
Merit: 10
Kinex - The New Frontier
LMFAO ...
well said ... the story of my life!
] http://articles.dailypress.com/2005-04-02/news/0504020090_1_marvin-p-smith-tour-guide-vertebrae]
Screw you guys - I was going home ...
Ski
sr. member
Activity: 440
Merit: 250
no refunds
no machines
and hashing with customers machines

this is realy crap

is worst than butterfly pufff
hero member
Activity: 728
Merit: 500
The next few difficulty adjustments are going to be brutal. I'm even more content now that I took the Neptupiter (Juptune?) offer and didn't wait for the Neptune. Still, I would have been much better off if I'd taken the refund, lol.

But now I'm absolutely certain I will never order another miner as long as I live. And I'll probably be switching over to some SHA shitcoins sooner than I thought.


I could say exactly the same!
full member
Activity: 131
Merit: 100
Here is a datasheet for a typical remote diode temperature management/fan control IC.

http://www.ti.com/lit/ds/symlink/lm96063.pdf

Note: "The LM96063 is remote diode temperature sensors with integrated fan control that includes remote diode sensing. The LM96063 accurately measures: (1) its own temperature and (2) the temperature of a diode- connected transistor, such as a 2N3904, or a thermal diode commonly found on Computer Processors, Graphics Processor Units (GPU) and other ASIC's."

These type of devices typically integrate with an on die thermal diode.

On the issue of BRM46x thermal design constraints, the spec says 125 C. Not 100 C, not 105 C. Looking at the data sheet, it looks like the device temperature is reported via the PMBus and may also be subject to configuration via a configuration file:

"This product is designed with a digital control circuit. The control circuit uses a configuration file which determines the functionality and performance of the product. The Electrical Specification table shows parameter values of functionality and performance with the default configuration file, unless otherwise specified. The default configuration file is designed to fit most application needs with focus on high efficiency. If different characteristics are required it is possible to change the configuration file to optimize certain performance characteristics. Note that current sharing operation requires changed configuration file. See application notes AN307 for further information."

Is anyone having problems with Neptune's shutting down due to thermal issues or is this yet more FUD being spread by haters and envious ne'er do wells towards KNC ?


Thanks for the information on the LM96063 and pointing out facts throughout this thread. I work with equipment where I can measure temperatures 100s of yards away so I don't understand why anyone would thing a small IC on the side of a board is not capable of reading temperatures directly inside an ASIC several inches away. The more I investigate and sift through the postings of people who don't really know the more comfortable I am feeling with the Neptune.

I have no doubt that a better heat sinking arrangement for the VRMs is in the works or could be made. At the least it's an easy thing to modify and I'll get my 12 year old (because of smaller fingers and  better eyes) to take care of it for me. Based on the specs I would like to see temperatures maintained below 100 and preferably 80 to 90 degrees C on the hottest VRM.
legendary
Activity: 1036
Merit: 1000
Nighty Night Don't Let The Trolls Bite Nom Nom Nom
pretty sure it was the weekly pool report that figured evwryth6ing out over the last while, they have slowly been figuring it all out.

I even helped them fill in a blank at one point a few months ago.

 http://organofcorti.blogspot.nl/?m=1
legendary
Activity: 3990
Merit: 4597
This shit is crazy.... It really looks like they are mining with our machines RIGHT NOW.
I can think of no other way that hashpower could possibly magically appear
https://blockchain.info/pools?timespan=48hrs
https://i.imgur.com/nZ1gpav.png

Looks like a BFL replay...   https://bitcointalksearch.org/topic/i-no-longer-work-for-butterfly-labs-bfl-659828
The drama of it all....  Holy Swashbucklers.


Kudos to the folks at blockchain.info for putting all that together and figuring out who was who. They've always been mining but they would show up as the Known Other which was close to 40% before they broke it out out recently. So, what it doesn't show is the hashrate that KnC might be using on those other pools if they are using that. However, considering how greedy those bastards are they probably don't want to share anything with some lowly pool. They have no image to protect nor do they care either.

KnC came from the banking industry and they behave like a bank with all that small print, email offers, etc.
For a bank your deposit is a liability (it is called that-surprisingly). Bank uses the deposit sum to make loans that are a multiple of your deposit, or buy 10Xtreasuries=40% per year for them. Same for KnC-your orders are KnC liabilities, so they use your preorder cash and order a multiple of Neptunes (chips), then they give you back as little as possible, while using the rest for themselves and by doing so, undermine your profit. Nice business model if you can maintain it-but you just CAN'T.
hero member
Activity: 744
Merit: 514
gotta let a coin be a coin
This shit is crazy.... It really looks like they are mining with our machines RIGHT NOW.
I can think of no other way that hashpower could possibly magically appear
https://blockchain.info/pools?timespan=48hrs


Looks like a BFL replay...   https://bitcointalksearch.org/topic/i-no-longer-work-for-butterfly-labs-bfl-659828
The drama of it all....  Holy Swashbucklers.


Kudos to the folks at blockchain.info for putting all that together and figuring out who was who. They've always been mining but they would show up as the Known Other which was close to 40% before they broke it out out recently. So, what it doesn't show is the hashrate that KnC might be using on those other pools if they are using that. However, considering how greedy those bastards are they probably don't want to share anything with some lowly pool. They have no image to protect nor do they care either.
hero member
Activity: 714
Merit: 500
This shit is crazy.... It really looks like they are mining with our machines RIGHT NOW.
I can think of no other way that hashpower could possibly magically appear

Looks like a BFL replay...   https://bitcointalksearch.org/topic/i-no-longer-work-for-butterfly-labs-bfl-659828
The drama of it all....  Holy Swashbucklers.

There is nothing magical about it.

They had atleast the same number of miners on order than the amount they presold.

As more than half of their customers refunded, they now have atleast 9PH ready to deploy.
hero member
Activity: 784
Merit: 504
Dream become broken often
i find it funny that everyone is so surprised knc did everyone like they did...can you blame them? could you keep greed monster at bay to be loyal to your customers? i doubt it...then ya say well they'll lose all their customers...don't think their worried about that when they now have their own farm and don't need us anymore...this will happen to all asic companies...greed monster always wins and customers always lose...soon it'll be centralized currency and then poof there goes bitcoin....and all the asic companies will shoot themselves in the foot...but not before making out like bandits so i guess they really won't care about that...imho

well it'll be great to see the ddos'ers go after knc now if they do...and if they don't well maybe its knc who is doing the attacks lmao
legendary
Activity: 1400
Merit: 1000
I owe my soul to the Bitcoin code...
Apologies if this was answered previously, I'm very far behind on this thread.

From what I've read a Neptune is 5 boxes and draws 2100 watts and each box has 1 PCI-e 6-pin power cable, that means that each single power cable is 420 watts, when a 6-pin is rated at 75 watts.  I know I've been running a November board and October boards at 190 to 200 watts on a single cable, and they get warm.... is there some mistake about the 1 PCI-e cable to power a 420 watt chip???

Technically the Neptune's boxes draw a total of ~1900W if you factor in AC/DC efficiency.  Molex's PCI-E keyed parts are rated for 8A per pin, so ~24A.  Which, at 12V gives you a rated operational power level of 288W.  However, in reality when you load things 24/7 you should de-rate them by 80%, so really any competent EE would not have these connectors drawing over ~230W (~19A @ 12V) in this type of application.  Now, we don't know the part number of the plug that KnC used, and some of Molex's Mini-fit Jr plugs handle up to 13A a pin which would allow ~468W running at their rating.  De-rate that and you're at ~375W which is just shy of each cubes ~380W draw.

TL;DR: There's 1 PCI-E cable per box, so yes: shitty Engineering.

Also most psu 6/8 pin pci-e wires are 18awg instead of 16awg.

I had a gander at the minifit jr. product specifications and even for the HCS pins/connnectors the max is only 13A for single row housings.  When you make double row housings like on a PCIe connector (2x3) the rated current goes down to 11A on 16AWG wiring.  Less so for smaller gauge wire.    Found here: http://www.molex.com/pdm_docs/ps/PS-45750-001.pdf

I just don't think I would want to run that out of spec 24/7/365.
hero member
Activity: 561
Merit: 521
Trustless IceColdWallet
no others got an neptun here?

elenenen =orama?
hero member
Activity: 882
Merit: 1003
Gonna be biggest diff jump yet?

There was a 46% jump in October
hero member
Activity: 910
Merit: 1004
buy silver!
Gonna be biggest diff jump yet?
legendary
Activity: 938
Merit: 1000
LIR DEV
This shit is crazy.... It really looks like they are mining with our machines RIGHT NOW.
I can think of no other way that hashpower could possibly magically appear
https://blockchain.info/pools?timespan=48hrs


Looks like a BFL replay...   https://bitcointalksearch.org/topic/i-no-longer-work-for-butterfly-labs-bfl-659828
The drama of it all....  Holy Swashbucklers.
hero member
Activity: 882
Merit: 1003
Apologies if this was answered previously, I'm very far behind on this thread.

From what I've read a Neptune is 5 boxes and draws 2100 watts and each box has 1 PCI-e 6-pin power cable, that means that each single power cable is 420 watts, when a 6-pin is rated at 75 watts.  I know I've been running a November board and October boards at 190 to 200 watts on a single cable, and they get warm.... is there some mistake about the 1 PCI-e cable to power a 420 watt chip???

Technically the Neptune's boxes draw a total of ~1900W if you factor in AC/DC efficiency.  Molex's PCI-E keyed parts are rated for 8A per pin, so ~24A.  Which, at 12V gives you a rated operational power level of 288W.  However, in reality when you load things 24/7 you should de-rate them by 80%, so really any competent EE would not have these connectors drawing over ~230W (~19A @ 12V) in this type of application.  Now, we don't know the part number of the plug that KnC used, and some of Molex's Mini-fit Jr plugs handle up to 13A a pin which would allow ~468W running at their rating.  De-rate that and you're at ~375W which is just shy of each cubes ~380W draw.

TL;DR: There's 1 PCI-E cable per box, so yes: shitty Engineering.

Also most psu 6/8 pin pci-e wires are 18awg instead of 16awg.
hero member
Activity: 714
Merit: 500
Apologies if this was answered previously, I'm very far behind on this thread.

From what I've read a Neptune is 5 boxes and draws 2100 watts and each box has 1 PCI-e 6-pin power cable, that means that each single power cable is 420 watts, when a 6-pin is rated at 75 watts.  I know I've been running a November board and October boards at 190 to 200 watts on a single cable, and they get warm.... is there some mistake about the 1 PCI-e cable to power a 420 watt chip???

Technically the Neptune's boxes draw a total of ~1900W if you factor in AC/DC efficiency.  Molex's PCI-E keyed parts are rated for 8A per pin, so ~24A.  Which, at 12V gives you a rated operational power level of 288W.  However, in reality when you load things 24/7 you should de-rate them by 80%, so really any competent EE would not have these connectors drawing over ~230W (~19A @ 12V) in this type of application.  Now, we don't know the part number of the plug that KnC used, and some of Molex's Mini-fit Jr plugs handle up to 13A a pin which would allow ~468W running at their rating.  De-rate that and you're at ~375W which is just shy of each cubes ~380W draw.

TL;DR: There's 1 PCI-E cable per box, so yes: shitty Engineering.
They just don´t give a shit, as their main deployment of these devices will be their DCs.

With an ambient air temperature of ~0°C-15°C (depending on time of year) they are quite sure that THEIR connectors won´t overheat.

Now, i would be careful to run these at 30-35°C ambient temperature, as the wires will more easily start electrical fires then.
legendary
Activity: 1512
Merit: 1000
Apologies if this was answered previously, I'm very far behind on this thread.

From what I've read a Neptune is 5 boxes and draws 2100 watts and each box has 1 PCI-e 6-pin power cable, that means that each single power cable is 420 watts, when a 6-pin is rated at 75 watts.  I know I've been running a November board and October boards at 190 to 200 watts on a single cable, and they get warm.... is there some mistake about the 1 PCI-e cable to power a 420 watt chip???

Technically the Neptune's boxes draw a total of ~1900W if you factor in AC/DC efficiency.  Molex's PCI-E keyed parts are rated for 8A per pin, so ~24A.  Which, at 12V gives you a rated operational power level of 288W.  However, in reality when you load things 24/7 you should de-rate them by 80%, so really any competent EE would not have these connectors drawing over ~230W (~19A @ 12V) in this type of application.  Now, we don't know the part number of the plug that KnC used, and some of Molex's Mini-fit Jr plugs handle up to 13A a pin which would allow ~468W running at their rating.  De-rate that and you're at ~375W which is just shy of each cubes ~380W draw.

TL;DR: There's 1 PCI-E cable per box, so yes: shitty Engineering.
Jump to: