Swedish ASIC miner company kncminer.com

[r1senfa17h]

Hey,

Where has "The Avenger" been lately? I haven't seen him posting since HashFast went into its bunker. Coincidence?

</runs away after kicking hornet's nest>

[1714226728]

1714226728

[1714226728]

1714226728

[1714226728]

1714226728

[1714226728]

1714226728

[The Avenger]

Hey,

Where has "The Avenger" been lately? I haven't seen him posting since HashFast went into its bunker. Coincidence?

</runs away after kicking hornet's nest>

There is nothing to talk about which I haven't said already, i.e. knc ship more Jupiters - will they won't they?

... KnC knows it has the Lexus of Bitcoin Miners at present.  KnC can sell at will all it offers.

They can't. As I've been saying, if they flood the market with new Jupiters, it'll massively impact neptune sales/refunds. Anyone with money would buy a dozen jupiters for delivery in Jan/Feb and cancel all neptune orders. Because 3TH now is worth XXTH in June/July. They can't risk a mass refund on neptunes, as they will have no business. Also, if they flood the market with Jupiters, it'll increase the network hashrate dramatically and that'll make the neptune look less and less attractive. Could even kill it.

So, they are in a delightfully difficult situation   KNC can

1. Sell loads of Jupiters and forget neptune. Lot's of money, but the end of their business.
2. Sell limited number of Jupiters and hope it doesn't damage neptune sales or lead to mass neptune refunds (most likely scenario).
3. Sell loads of Jupiters, but make sure people with neptune orders cannot get a refund. In the process, make themselves the most hated asic company in the world.
4. Don't sell any new Jupiters and rely on neptune. But this bet relies on their competition screwing up and not being able to take the lead for the X months it takes knc to squeeze out the neptune. If competion surges ahead, knc are left behind. Any major delays in neptune - knc are dead.

How many more options do they have?

Ok, if KnC announces a new batch of Jupiters - you can thank me. Why? Because I just spent all of my bitcoins on 13 x 200GH/s Avalon clones that are shipping "immediately" (haha, we'll see). To make me regret my decision, KnC will almost definitely sell the Jupiters I want now. You're welcome.

P.S. I thought avalon clones were only pre-order right now and models were 20-60GH?

[Xian01]

Also can you show me which GPU "pull in excess of that"?

 My bad. Was looking at dual gpu figures on tomshardware. 6990's and 7990's are beasts, for example.

EDIT: ... and FWIW, just pulled the trigger on a batch 2 Neptune order. Whee !
 
*holds everyone tightly*

[vesperwillow]

What GPU pulls 350W?

I run 7970/280X decently overclocked. They each pull 340-360 on a good day. More if you want to really squash your balls on the wall, I've had them pull 380 before.

KNC board uses 3x 12V circuits to carry up to 15A of current.

A GPU uses 6x 12V circuits to carry up to 23A of current.

Which one looks more to you?

Yes, I understand 3x5a=15a and 6x3.8a=23a.

Are you aware that the PCI-E slot provides 75W of the total power?
Are you using powered PCI-E risers?

When you answer these questions you will understand that each 12V circuit on the GPU cabling uses less that the KNC board

Yeah I'm aware of PCB design, and how the PCIx platform provides power. No, I don't use powered risers, all of the juice flows through the PCIE connectors, which means they flow more current than a normal setup.

You also should follow the conversation, specifically my ignored my question regarding pushing the KNC clock beyond stock to cause the wires to reach danger levels. Most 18AWG has a 4x-6x safety rating built in (depending on purity, oxygen content and sleeving)--even at 10amps they should be safe bundled together. Bundled @ 5amps (KNC design) they shouldn't be a danger.

KNC isn't loopy, which is why they knew they could use 18AWG for the boards.

This is why I proposed some airflow across them for anyone concerned about temps, and that they shouldn't be reaching such high temps anyhow unless something is wrong. Again, mine don't get near that, another reason I was asking about overclock.

Edit: My modules are running 145-155gh, the cable temps are 36.9C, the room is 26.7C. I have no form of cooling on the cables.

[r1senfa17h]

Hey,

Where has "The Avenger" been lately? I haven't seen him posting since HashFast went into its bunker. Coincidence?

</runs away after kicking hornet's nest>

There is nothing to talk about which I haven't said already, i.e. knc ship more Jupiters - will they won't they?

... KnC knows it has the Lexus of Bitcoin Miners at present.  KnC can sell at will all it offers.

They can't. As I've been saying, if they flood the market with new Jupiters, it'll massively impact neptune sales/refunds. Anyone with money would buy a dozen jupiters for delivery in Jan/Feb and cancel all neptune orders. Because 3TH now is worth XXTH in June/July. They can't risk a mass refund on neptunes, as they will have no business. Also, if they flood the market with Jupiters, it'll increase the network hashrate dramatically and that'll make the neptune look less and less attractive. Could even kill it.

So, they are in a delightfully difficult situation   KNC can

1. Sell loads of Jupiters and forget neptune. Lot's of money, but the end of their business.
2. Sell limited number of Jupiters and hope it doesn't damage neptune sales or lead to mass neptune refunds (most likely scenario).
3. Sell loads of Jupiters, but make sure people with neptune orders cannot get a refund. In the process, make themselves the most hated asic company in the world.
4. Don't sell any new Jupiters and rely on neptune. But this bet relies on their competition screwing up and not being able to take the lead for the X months it takes knc to squeeze out the neptune. If competion surges ahead, knc are left behind. Any major delays in neptune - knc are dead.

How many more options do they have?

Ok, if KnC announces a new batch of Jupiters - you can thank me. Why? Because I just spent all of my bitcoins on 13 x 200GH/s Avalon clones that are shipping "immediately" (haha, we'll see). To make me regret my decision, KnC will almost definitely sell the Jupiters I want now. You're welcome.

P.S. I though avalon clones were only pre-order right now and models were 20-60GH?

My bad about your absence, sorry. Here's the thread with the Avalon BTMine units I was referring to.

[seanrarey]

I have GPUs pulling 350-360watts (29-30amps). I wasn't under the impression that the KNC modules were pulling that. I could be mistaken.

You surely must be mistaken:

What GPU pulls 350W?
How many cables are you using per GPU?
Are you aware that the PCI-E slot provides 75W of the total power?
Are you using powered PCI-E risers?

When you answer these questions you will understand that each 12V circuit on the GPU cabling uses less that the KNC board

I run banks of 7970 GPUs pulling up to 450w _EACH_... yeah, it adds up.

[The Avenger]

My bad about your absence, sorry. Here's the thread with the Avalon BTMine units I was referring to.

btmine only set up their bitcointalk username in mid-December?? 

It's a hella risky investment if you've gone all-in. But the factory photo's do look encouraging. Good luck with it.


The only thing I'd add to the ongoing jupiter speculation is the fact byteminr need new hashing power. And they've said on their blog they expect it in January 2014.

And if you know who's who, you know they will only get that hashing power from knc. They are most likely only paying cost (or less) for this hashing power (if you know who's who, that's obvious), so no way they'll be getting it from anyone else.

So the real cost of production would need to be taken up by people paying the full marked-up Jupiter price. I think this was the plan back in December when knc said have some coins ready. The timing would certainly fit to have a production run in January. But perhaps knc hadn't fully considered the impact on neptune and their future business if they did that?

The question is - can knc do a private sale of X jupiters to cover the production costs for byteminr units, which means they would not have to open it up to all the hungry miners in this thread?

Or will they open it up to everyone, but a limited sale? Or maybe another sale of boards? Or maybe they sell all the October Jupiters they have in the data centre and replace them with the cleaner November design? 

Who knows - all speculation at present. It'll be very interesting to see what they do.

[kmacri]

i've got a saturn and a jupiter.  scored them big local on my cl.  think i'll flip the saturn though.  i would prefer dual jups...

[faetos]

Also can you show me which GPU "pull in excess of that"?

 My bad. Was looking at dual gpu figures on tomshardware. 6990's and 7990's are beasts, for example.

EDIT: ... and FWIW, just pulled the trigger on a batch 2 Neptune order. Whee !
 
*holds everyone tightly*

*Big Hugz!*

[Paladin69]

It doesn't make sense why they would.  They are hurting their own cash cow too otherwise.

Not if they think they can get away with it. For the record, there is already very compelling evidence ghash.io was abused to pull off double spend attacks:
https://bitcointalk.org/index.php?topic=327767.0

Now you may not have sympathy for the victim here,  being a gambling site accepting zero confirmation transactions, but if this is not reason enough for you to switch pools, honestly you deserve what it could lead to.

Playing devil's advocate more than anything.  I have been on BTC Guild for a while now.

[CYPER]

What GPU pulls 350W?

I run 7970/280X decently overclocked. They each pull 340-360 on a good day. More if you want to really squash your balls on the wall, I've had them pull 380 before.

KNC board uses 3x 12V circuits to carry up to 15A of current.

A GPU uses 6x 12V circuits to carry up to 23A of current.

Which one looks more to you?

Yes, I understand 3x5a=15a and 6x3.8a=23a.

Are you aware that the PCI-E slot provides 75W of the total power?
Are you using powered PCI-E risers?

When you answer these questions you will understand that each 12V circuit on the GPU cabling uses less that the KNC board

Yeah I'm aware of PCB design, and how the PCIx platform provides power. No, I don't use powered risers, all of the juice flows through the PCIE connectors, which means they flow more current than a normal setup.

You also should follow the conversation, specifically my ignored my question regarding pushing the KNC clock beyond stock to cause the wires to reach danger levels. Most 18AWG has a 4x-6x safety rating built in (depending on purity, oxygen content and sleeving)--even at 10amps they should be safe bundled together. Bundled @ 5amps (KNC design) they shouldn't be a danger.

KNC isn't loopy, which is why they knew they could use 18AWG for the boards.

This is why I proposed some airflow across them for anyone concerned about temps, and that they shouldn't be reaching such high temps anyhow unless something is wrong. Again, mine don't get near that, another reason I was asking about overclock.

Edit: My modules are running 145-155gh, the cable temps are 36.9C, the room is 26.7C. I have no form of cooling on the cables.

The bottom line is no GPU card can pull as much current per cable as a KNC board does.

I run banks of 7970 GPUs pulling up to 450w _EACH_... yeah, it adds up.

I highly doubt that. I have a 7970 myself, so can speak from experience. I've ran benchmarks @ 1340Mhz / 7600Mhz and 1.33V core which is a lot.

[vesperwillow]

The bottom line is no GPU card can pull as much current per cable as a KNC board does.

...

I highly doubt that. I have a 7970 myself, so can speak from experience. I've ran benchmarks @ 1340Mhz / 7600Mhz and 1.33V core which is a lot.

While I would agree a KNC module *per cable* will likely draw more current, I was originally pointing out I find it odd for them to be getting as warm as they are, ie, something must be up with the KNC module if it's not overclocked but drawing excessive amounts of current. I guess it could just be component variance.

Your experience, like mine, doesn't dictate everyone's. I've seen individual GPUs past 400, I've pushed mine to 380ish. I wouldn't find it impossible to see one in the mid 400's, just not the normal case. With the right cooling, MSI cards can take 3 hells' worth of beatings, as I'm sure some other cards can, especially when some of the cores start at 1.2905.

[Biffa]

From what I understand, 51%'ing a network does no harm unless the pool operator has bad intentions.

Why risk it though? What does ghash.io give you that BTCGuild or eligius doesnt?

[ImI]

From what I understand, 51%'ing a network does no harm unless the pool operator has bad intentions.

Why risk it though? What does ghash.io give you that BTCGuild or eligius doesnt?

BTCGuild = 3% Fee (Thats ALOT when it comes to bigger BTC amounts)
Eligius = due to some problems i seem to get there around 2% less than ghash.io

So it DOES matter where you mine.

Nevertheless it should be obv that its time to mine somewhere else regardless the performance.

[CYPER]

From what I understand, 51%'ing a network does no harm unless the pool operator has bad intentions.

Why risk it though? What does ghash.io give you that BTCGuild or eligius doesnt?

Compared to BTCGuild: 0% fees
Compared to Eligius: less variance

I started mining with them around 3am UK time on 03.01.2014, so in about 1 hour that will be exactly 1 week mining.
My average hashrate is 3200GH/s.
For that 1 week period I have mined 8.54603089BTC exactly (not counting the last 1 hour of the full 1 week), while the calculator says 7.94166635 BTC per week.
So (8.54603089/7.94166635)*100 = 107.6% luck
So all looks good.

[Biffa]

From what I understand, 51%'ing a network does no harm unless the pool operator has bad intentions.

Why risk it though? What does ghash.io give you that BTCGuild or eligius doesnt?

Compared to BTCGuild: 0% fees
Compared to Eligius: less variance

I started mining with them around 3am UK time on 03.01.2014, so in about 1 hour that will be exactly 1 week mining.
My average hashrate is 3200GH/s.
For that 1 week period I have mined 8.54603089BTC exactly (not counting the last 1 hour of the full 1 week), while the calculator says 7.94166635 BTC per week.
So (8.54603089/7.94166635)*100 = 107.6% luck
So all looks good.

Forgot about BTCGuilds fee Now I remember why I don't use them.

But Eligius has been fine for me, I don't have your 6 Jupiters worth of hash rate but its been pretty good on there for me compared to slush.

I've steered clear of ghash.io for ages because they were just too big.

[seanrarey]

What GPU pulls 350W?

I run 7970/280X decently overclocked. They each pull 340-360 on a good day. More if you want to really squash your balls on the wall, I've had them pull 380 before.

KNC board uses 3x 12V circuits to carry up to 15A of current.

A GPU uses 6x 12V circuits to carry up to 23A of current.

Which one looks more to you?

Yes, I understand 3x5a=15a and 6x3.8a=23a.

Are you aware that the PCI-E slot provides 75W of the total power?
Are you using powered PCI-E risers?

When you answer these questions you will understand that each 12V circuit on the GPU cabling uses less that the KNC board

Yeah I'm aware of PCB design, and how the PCIx platform provides power. No, I don't use powered risers, all of the juice flows through the PCIE connectors, which means they flow more current than a normal setup.

You also should follow the conversation, specifically my ignored my question regarding pushing the KNC clock beyond stock to cause the wires to reach danger levels. Most 18AWG has a 4x-6x safety rating built in (depending on purity, oxygen content and sleeving)--even at 10amps they should be safe bundled together. Bundled @ 5amps (KNC design) they shouldn't be a danger.

KNC isn't loopy, which is why they knew they could use 18AWG for the boards.

This is why I proposed some airflow across them for anyone concerned about temps, and that they shouldn't be reaching such high temps anyhow unless something is wrong. Again, mine don't get near that, another reason I was asking about overclock.

Edit: My modules are running 145-155gh, the cable temps are 36.9C, the room is 26.7C. I have no form of cooling on the cables.

The bottom line is no GPU card can pull as much current per cable as a KNC board does.

I run banks of 7970 GPUs pulling up to 450w _EACH_... yeah, it adds up.

I highly doubt that. I have a 7970 myself, so can speak from experience. I've ran benchmarks @ 1340Mhz / 7600Mhz and 1.33V core which is a lot.

Wow.

Mr Cypher, any experienced GPU miner will tell you that a tacked 7970 will pull more then 400watts.  I tend to a substantial GPU farm on an hourly basis (it now mines LTC). 


... so you are simply wrong.  move on.

[merv77]

From what I understand, 51%'ing a network does no harm unless the pool operator has bad intentions.

Why risk it though? What does ghash.io give you that BTCGuild or eligius doesnt?

Compared to BTCGuild: 0% fees
Compared to Eligius: less variance

I started mining with them around 3am UK time on 03.01.2014, so in about 1 hour that will be exactly 1 week mining.
My average hashrate is 3200GH/s.
For that 1 week period I have mined 8.54603089BTC exactly (not counting the last 1 hour of the full 1 week), while the calculator says 7.94166635 BTC per week.
So (8.54603089/7.94166635)*100 = 107.6% luck
So all looks good.

that's a bit greedy.. you'd rather risk the bitcoin network and go against Shatoshi Nakamoto's design just for a little extra coins.  

even ghash.io has admitted that they are a threat to the bitcoin network and will make good... but still there are a few people that just don't care

[CYPER]

From what I understand, 51%'ing a network does no harm unless the pool operator has bad intentions.

Why risk it though? What does ghash.io give you that BTCGuild or eligius doesnt?

Compared to BTCGuild: 0% fees
Compared to Eligius: less variance

I started mining with them around 3am UK time on 03.01.2014, so in about 1 hour that will be exactly 1 week mining.
My average hashrate is 3200GH/s.
For that 1 week period I have mined 8.54603089BTC exactly (not counting the last 1 hour of the full 1 week), while the calculator says 7.94166635 BTC per week.
So (8.54603089/7.94166635)*100 = 107.6% luck
So all looks good.

that's a bit greedy.. you'd rather risk the bitcoin network and go against Shatoshi Nakamoto's design just for a little extra coins.  

even ghash.io has admitted that they are a threat to the bitcoin network and will make good... but still there are a few people that just don't care

Ghash.io is far from having 50% of the network, but yet people freak out.
Obviously it is impossible to predict their exact share as this is based on previously solved blocks and luck, but roughly it is around 40% currently, which is far from 50%.

According to http://bitcoincharts.com/markets/ total hashrate is close to 16PH/s
Ghash.io has 4.7PH/s
That's less than 30%

If they get really close to 50% then I will surely move out.

[DPoS]

If they get really close to 50% then I will surely move out.

all Hail cyper for saving us