Swedish ASIC miner company kncminer.com - page 1175.

vesperwillow

hero member

Activity: 616

Merit: 500

Quote from: DPoS on October 31, 2013, 11:45:46 AM

Quote from: vesperwillow on October 31, 2013, 11:11:48 AM

hot datacenters

I guess the crappy ones that couldn't stay cool have a new selling point...

Not all of them have the same average temperature, and the overengineering on KNC's part allows for a greater fluctuation without impacting the hash potential. Plus, some folks have lots of computing power but not adequate cooling.. so it works well.

Either way.. totally called it Wink

DPoS

sr. member

Activity: 462

Merit: 250

Quote from: vesperwillow on October 31, 2013, 11:11:48 AM

hot datacenters

I guess the crappy ones that couldn't stay cool have a new selling point...

vesperwillow

hero member

Activity: 616

Merit: 500

Quote from: FiatKiller on October 31, 2013, 10:35:15 AM

Quote from: vesperwillow on October 31, 2013, 10:18:20 AM

For those noticing that some of their units need to be warmer for better rates, I have no doubt about that. I have the same issue with my GPU farm of MSI TF 7950's. Some of the cards don't like being cold at all. The cooler they get, the higher the error rate and the lower the hash. The warmer, the inverse occurs.

For these cards, the sweet spot seems to be 75-82. Error rate is less than a percent, hash rate is easily 690+. Below 75 they slow down and error rate doubles or even triples.

It's the 28nm architecture.

Interesting, a chart on this page shows that 28 nm comparator sensitivity(perhaps to EMF?) drops and levels-off
at the temps we've noticed working well:

http://www.design-reuse.com/articles/28591/mixed-signal-ip-design-challenges-in-28-nm-and-beyond.html

Neat link.

And though way earlier in this thread I was dogged for not being an electrical component design engineer while making some comments on design, assembly and deployment.. every now and then, it's as if I know what I'm talking about. lol

I predicted the 28nm would have issues if too cold. Looks like I'm partially right. One thing is for sure, they designed the cooling setup to work in hot datacenters, so it's no wonder some of these units aren't able to warm-up when used in houses.

FiatKiller

sr. member

Activity: 378

Merit: 250

Quote from: vesperwillow on October 31, 2013, 10:18:20 AM

For those noticing that some of their units need to be warmer for better rates, I have no doubt about that. I have the same issue with my GPU farm of MSI TF 7950's. Some of the cards don't like being cold at all. The cooler they get, the higher the error rate and the lower the hash. The warmer, the inverse occurs.

For these cards, the sweet spot seems to be 75-82. Error rate is less than a percent, hash rate is easily 690+. Below 75 they slow down and error rate doubles or even triples.

It's the 28nm architecture.

Interesting, a chart on this page shows that 28 nm comparator sensitivity(perhaps to EMF?) drops and levels-off
at the temps we've noticed working well:

http://www.design-reuse.com/articles/28591/mixed-signal-ip-design-challenges-in-28-nm-and-beyond.html

vesperwillow

hero member

Activity: 616

Merit: 500

For those noticing that some of their units need to be warmer for better rates, I have no doubt about that. I have the same issue with my GPU farm of MSI TF 7950's. Some of the cards don't like being cold at all. The cooler they get, the higher the error rate and the lower the hash. The warmer, the inverse occurs.

For these cards, the sweet spot seems to be 75-82. Error rate is less than a percent, hash rate is easily 690+. Below 75 they slow down and error rate doubles or even triples.

It's the 28nm architecture.

DobZombie

hero member

Activity: 896

Merit: 532

Former curator of The Bitcoin Museum

I'm only getting 93-99GH/s on my Mercury Sad

Tried 0.98
tried 0.96
tried enablecores.bin on both
tried different pools

*sighs*

Xialla

legendary

Activity: 1036

Merit: 1001

/dev/null

Quote from: ImI on October 31, 2013, 08:41:17 AM

Quote from: Xialla on October 31, 2013, 08:26:32 AM

.98 after ~28hours

nice! any special treatments? extra fans etc?

nope, just opened window in the room, case on.

ASIC slot #1   55.0 ℃
ASIC slot #2   58.0 ℃
ASIC slot #3   61.5 ℃
ASIC slot #4   56.0 ℃

..from now, I'm scared to change anything.)

ImI

legendary

Activity: 1946

Merit: 1019

Quote from: Xialla on October 31, 2013, 08:26:32 AM

.98 after ~28hours

nice! any special treatments? extra fans etc?

FiatKiller

sr. member

Activity: 378

Merit: 250

Quote from: Madazam on October 31, 2013, 08:18:17 AM

Quote from: FiatKiller on October 31, 2013, 07:56:59 AM

Ok, been running over a day now no problem:

Is that an experiment ? why the plastic bags ....?

to raise the temperature closer to ideal in my cold basement

Biffa

legendary

Activity: 3234

Merit: 1221

Found my Jupiter hashing at 140Ghs just now (cgminer and pool), so back to 0.96 for me.

Xialla

legendary

Activity: 1036

Merit: 1001

/dev/null

.98 after ~28hours

madsusies

sr. member

Activity: 428

Merit: 250

BitcoinEvo [$XBTE]

Lil awes pic, meanwhile it looks like the mercury earn a 26€ daiy hard reality,i can buy a bottle of wine

Madazam

member

Activity: 115

Merit: 10

Quote from: FiatKiller on October 31, 2013, 07:56:59 AM

Ok, been running over a day now no problem:

Is that an experiment ? why the plastic bags ....?

sickpig

legendary

Activity: 1260

Merit: 1008

Quote from: ?? on ??

Here are my findings: https://docs.google.com/spreadsheet/...nc&usp=sharing

broken link

FiatKiller

sr. member

Activity: 378

Merit: 250

Ok, been running over a day now no problem: (reading 294 at the pool too!)

Mota

legendary

Activity: 804

Merit: 1002

Quote from: timmmers on October 31, 2013, 06:19:02 AM

Quote from: Mota on October 31, 2013, 04:35:24 AM

Quote from: timmmers on October 31, 2013, 04:26:51 AM

Quote from: DigginDeep on October 31, 2013, 02:20:34 AM

They told me they planned to have
A month in between shipping, and I believe they will hold up to that, other wise they would piss off allot of customers, and anyone smart that paid with a credit card could get a refund if they wanted, and would lose allot of customers if they plan on having a gen 2, though a gen 2 would only be worth while if they got much better watt / gig performance

Yeah, their shipping times have always been as they say, and they hate to piss customers off...or even talk to us come to that.

I'd bet that Nov orders are quite small given the difficulty we all knew was going to be this bad, and if power usage being lower than current rigs is such a big thing. Gen 2 will have to have a much better dollar / gig performance, who cares about power consumption of a few quid when you're talking about rigs that should be making hundreds a week? Power use should be about those last few months just above break even not massive thing. People see a 2 week delay as OK despite that meaning missing so much income, then witter on about the cost of a couple of light bulbs in power consumption? WTF?

I beg to differ. 1 Jupiter costs about 135€ in power per month. That is at present pretty much 1 btc. If next gen miners can lessen that by 30% that would mean a LOT more money in the long run. Of course, if you are in a country where you don't have to pay more than 13 cents per kwh you obviously don't have to fret over power consumption...

A jupiter would cover that 135 in 2 days sitting instead of waiting to be shipped at KNC. A next gen rig would need to be at least the same value as a jupiter right now to be even considered. Power is the last thing that needs to be improved, and any improvements will be pitiful anyway. IF the people who only just got their jupiters had been offered the choice of one on the 15th as promised but 1kw ...or what they got...14 days of no mining and a diff increase...which would they take? 2 weeks mining at 500GH or save 70 euros? Which is more important?

1BTC is a lot more than 135 Euro right now too.

Hey, don't preach to the choir. My UPS delivery from KNC directly is missing... so, 2 weeks late vs now 4 weeks with me.... I am lucky that my first units got here on the 16th late afternoon, also UPS, delivered to my neighbours babysitter (while I was at home...). But that was not the unit from KNC directly but from a forumbuy. So yeah, delay sucks, but when you get your units what is more important is power consumption.

timmmers

sr. member

Activity: 1176

Merit: 265

Quote from: Mota on October 31, 2013, 04:35:24 AM

Quote from: timmmers on October 31, 2013, 04:26:51 AM

Quote from: DigginDeep on October 31, 2013, 02:20:34 AM

They told me they planned to have
A month in between shipping, and I believe they will hold up to that, other wise they would piss off allot of customers, and anyone smart that paid with a credit card could get a refund if they wanted, and would lose allot of customers if they plan on having a gen 2, though a gen 2 would only be worth while if they got much better watt / gig performance

Yeah, their shipping times have always been as they say, and they hate to piss customers off...or even talk to us come to that.

I'd bet that Nov orders are quite small given the difficulty we all knew was going to be this bad, and if power usage being lower than current rigs is such a big thing. Gen 2 will have to have a much better dollar / gig performance, who cares about power consumption of a few quid when you're talking about rigs that should be making hundreds a week? Power use should be about those last few months just above break even not massive thing. People see a 2 week delay as OK despite that meaning missing so much income, then witter on about the cost of a couple of light bulbs in power consumption? WTF?

I beg to differ. 1 Jupiter costs about 135€ in power per month. That is at present pretty much 1 btc. If next gen miners can lessen that by 30% that would mean a LOT more money in the long run. Of course, if you are in a country where you don't have to pay more than 13 cents per kwh you obviously don't have to fret over power consumption...

A jupiter would cover that 135 in 2 days sitting instead of waiting to be shipped at KNC. A next gen rig would need to be at least the same value as a jupiter right now to be even considered. Power is the last thing that needs to be improved, and any improvements will be pitiful anyway. IF the people who only just got their jupiters had been offered the choice of one on the 15th as promised but 1kw ...or what they got...14 days of no mining and a diff increase...which would they take? 2 weeks mining at 500GH or save 70 euros? Which is more important?

1BTC is a lot more than 135 Euro right now too.

bronan

hero member

Activity: 774

Merit: 500

Lazy Lurker Reads Alot

Your right madsusie no miner is going to make roi unless a person is not paying for his electra bills then that person can make profit,
unless the value of btc becomes at a massive value to make up the difficulty.

Which i believe will not happen anytime soon.

However since i do have to pay 0.32 Kw/h on electra and pay massive taxes over anything i buy in my country.

Its not longer possible to make mining worth the effort. Colocating gives other problems in general makes things worse.
The problem with hosting is the hands when it goes wrong, and you need a person which has knowledge on these machines.

Second the prices which the asic companies ask for their $8 value chips is insane.
Do not come with bs about they cost more because they don't, their cheap and since the foundries have a new market they probably dropped even lower in price.

Asics are made in massive numbers each day since they are used in many products todo their respected tasks.
They are simple chips compared to cpu/gpu and not programmable like fpga so after this task you can not use asics at any other task which they are made for. They are instant useless so selling you expenssive bitcoin miner becomes a pain when its no longer able to make some coins for you.

madsusies

sr. member

Activity: 428

Merit: 250

BitcoinEvo [$XBTE]

100BTc=20k$ Cheesy

let my see, diff will be 1000mil min 2Thas not enough 20Thas to have a ROI unit with a lot consumption, literally there will be not roi units, but folks somehow don't care

capa

full member

Activity: 138

Merit: 100

Hi,

As many of you are already aware, there have been a lot of issues with firmware for the KnC series miners, attempts to replace this code with familiar tried and tested systems fails because
the system right now works in flash, which is why you loose it on reboot.

So my plan is to source a beaglebone and develop a microsd image with a more common OS like Debian or Ubuntu with BFGminer on that i can release to everyone thus surviving reboots.
If you like the idea donate here: 1Nm8mRuuCYQqDwY8HhNGPsAQ1KKE3ykQ7o

This will be a plug a play type setup with minimal config required.

I.e copy the image.iso to a microsd plug the microsd into your miner and reboot..

Smile

I've started a thread here https://bitcointalksearch.org/topic/knc-miner-custom-os-for-beaglebone-321930
And here http://forum.kncminer.com/blogs/the-big-countdown/13322-custom-os-image-for-knc-miners
and here http://abitcoin.org/viewtopic.php?f=11&t=780

Topic: Swedish ASIC miner company kncminer.com - page 1175. (Read 3049528 times)