Swedish ASIC miner company kncminer.com

[crashoveride54902]

I don't know how to post images, I am familiar with pastebin sites but not sure how people paste pictures.

The page shows wattages usually 265 to 297 or so, but flicks to tiny numbers like 3 or 9 watts from time to time for moments too.

Temperatures of dies in one ASIC range from 77 through 93 degrees. Possibly it is throttling a lot due to temperatures or something like that.

Actually just saw one have a range of 74 through 95.

It doesn't seem to be throttling the 95 either so maybe throttling due to temperature is not what is going on.

-MarkM-

think its your psu's...what kind are they? what watts? what pool ya using? you sure you have stable net to controller board? sounds like its going to idle cause either it is losing net or pool isn't sending work or i heard on IRC the BBB isn't the best...think its the black version and has problems...idk I looked back at elen pics of stock unit adv. tab and it showed 280w per asic or so...

just saw yer post...ya i'd look at psu or pool or controller board...do you have all the cables seated right as well?

[markm]

I have a 1000 watt Cougar CMX powering one cube and the controller, an Aurum AU-1000PRO powering one cube, two Cooler Master 725's each powering one cube, and a Cooler Master ZPG-1200B powering one cube. That is, five power supplies currently to power the five cubes plus one controller. Because one ZPG-1200B died trying to power three cubes and the other didn't seem to really suffice for two cubes (both fresh out of the box new from Newegg).

Network should not be a problem as I have a backup pool configured and both pools use stratum so even though I have more hash power the pools should simply adjust the share-difficulty to keep the net usage much the same. All cables seem fine. Besides if they were not they'd all have to be equally unfine wouldn't they since there seem to be no problem cubes just a general overall unexpectedly low hash rate for the whole kit and kaboodle.

-MarkM-

[Ski72]

Alright, I have finished a process for making a heavy duty 2:1 PCIe adapter for Neptune owners.  It is made with 16AWG wires and minifit HCS (high current system) pins capable of 13A.  It is tricky to make with crimping two 16 gauge wires together but I soldered the pins to the wires after crimping for best contact.  I cannot guarantee that the nylon housings at the pcb won't melt so be sure there is some airflow there.

Here she is:


Pricing is $8 each with discounts for larger orders.  Shipping will be priority mail with tracking which is $6 for US and $24 for international.
Please PM if you are interested.  Thanks

Another solution (but a lot more DIY work):

if anyone is interested, I have plates that replace the mounting apparatus they used.  It will get the foam out of the way of the VRMs and allow more airflow over the PCB.

They'll come with the plates, screws, nuts and spacers for all 5 ASICs - .0675 BTC plus shipping (USPS Priority/Priority Express, your choice)
PM me ...
Ski

Sorry but they will not fit !.... you need an T-shape plate, and not an I-shape

true, it won't fit with the box on it.  You'd have to turn the Arctic radiator 90^o and it will be mounted in the same fashion as the Jupiter.  It looks like everyone is doing their best to make it work. 
Just sayin' ...

[crashoveride54902]

I have a 1000 watt Cougar CMX powering one cube and the controller, an Aurum AU-1000PRO powering one cube, two Cooler Master 725's each powering one cube, and a Cooler Master ZPG-1200B powering one cube. That is, five power supplies currently to power the five cubes plus one controller. Because one ZPG-1200B died trying to power three cubes and the other didn't seem to really suffice for two cubes (both fresh out of the box new from Newegg).

Network should not be a problem as I have a backup pool configured and both pools use stratum so even though I have more hash power the pools should simply adjust the share-difficulty to keep the net usage much the same. All cables seem fine. Besides if they were not they'd all have to be equally unfine wouldn't they since there seem to be no problem cubes just a general overall unexpectedly low hash rate for the whole kit and kaboodle.

-MarkM-

well then its either the psu's or the asics themselves...really no reason why you shouldn't be hashing at 3th/s right now...another guy on IRC is having troubles too...and he said with every restart it seems to be getting worse...good luck to ya man...maybe you can hook one unit up to controller and power it with the cougar 1k w psu and see how it hashes and just go at it one by one and see if you have a problem asic that is causing problems with the rest of'em...have no clue just trying to help

[proclivity]

I have a 1000 watt Cougar CMX powering one cube and the controller, an Aurum AU-1000PRO powering one cube, two Cooler Master 725's each powering one cube, and a Cooler Master ZPG-1200B powering one cube. That is, five power supplies currently to power the five cubes plus one controller. Because one ZPG-1200B died trying to power three cubes and the other didn't seem to really suffice for two cubes (both fresh out of the box new from Newegg).

Network should not be a problem as I have a backup pool configured and both pools use stratum so even though I have more hash power the pools should simply adjust the share-difficulty to keep the net usage much the same. All cables seem fine. Besides if they were not they'd all have to be equally unfine wouldn't they since there seem to be no problem cubes just a general overall unexpectedly low hash rate for the whole kit and kaboodle.

-MarkM-

Login using putty and you'll be albe to see a lot more about what is happening. If you switch to BFGMiner you'll also be able to look at the cores to see if they are turning off and on. If you don't know how to use putty, there are some good posts that you will find with a couple of Google searches.

[tolip_wen]

That last ASIC happened to get snapshotted just when it showed low wattage; the numbers fluctate, so shortly it again looked much like all the others.

-MarkM-

You got low current on
ASIC 4 DC/DC 4&5
ASIC6 DC/DC 0,1,4,5,6,7

They sleepin or at least groggy

YMMV

[tolip_wen]

Login using putty and you'll be albe to see a lot more about what is happening. If you switch to BFGMiner you'll also be able to look at the cores to see if they are turning off and on. If you don't know how to use putty, there are some good posts that you will find with a couple of Google searches.

What FW has bfgminer?
Where to get?

I have rc6 and it is sans bfgminer.
The GUI button does nothing.

Was gonna hand roll, I'll take precompiled though.

YMMV

[proclivity]

What FW has bfgminer?
Where to get?

I have rc6 and it is sans bfgminer.
The GUI button does nothing.

Was gonna hand roll, I'll take precompiled though.

YMMV

My neppy's arrive tomorrow.. Uh oh, no BFGMiner? This was so helpful for me on the Jups..! I might just have to roll my own as you said tolip

At least by logging in he can still see the error rates and, if they are high, see which die is generating the problem and tweak the voltage to see if it improves.

[tolip_wen]

What FW has bfgminer?
Where to get?

I have rc6 and it is sans bfgminer.
The GUI button does nothing.

Was gonna hand roll, I'll take precompiled though.

YMMV

My neppy's arrive tomorrow.. Uh oh, no BFGMiner? This was so helpful for me on the Jups..! I might just have to roll my own as you said tolip

At least by logging in he can still see the error rates and, if they are high, see which die is generating the problem and tweak the voltage to see if it improves.

I miss bfgminer and the awk script I use for showing core HW errors in a grid like on the ASIC.
Plus HW totals for each die.

bfgminer also has temp data without the damn GUI.

One advantage to hand roll is can set core HW timeout differently if choose.

Things to do, things to do............

[yuriygeorge]

Was working fine with one AX1200i, and I even connected 3 boxes + controller board. Now, enter 2nd PSU, put 2 boxes + controller board on first PSU, and then 3 boxes on the new PSU, and not having things going smoothly. doing the troubleshooting thing where I adjust slowly up and will see how it goes, but still doesn't seem promising. The dies keep going from low to high and oscillating the Wattage.

Is there a special order that we need to turn the 2 PSU's on? Controller board first? then Second PSU? Or at the same time? or one shortly after the other?

[crashoveride54902]

Was working fine with one AX1200i, and I even connected 3 boxes + controller board. Now, enter 2nd PSU, put 2 boxes + controller board on first PSU, and then 3 boxes on the new PSU, and not having things going smoothly. doing the troubleshooting thing where I adjust slowly up and will see how it goes, but still doesn't seem promising. The dies keep going from low to high and oscillating the Wattage.

Is there a special order that we need to turn the 2 PSU's on? Controller board first? then Second PSU? Or at the same time? or one shortly after the other?

hmm your not the only having this problem...maybe its the controller board...can't handle all 5 units? or the FW is the issue...who knows...good luck to you guys

[yuriygeorge]

It was working fantastic with just 1 PSU, and 3 boxes + 1 controller board, was getting over 2TH. Now when I introduce the 2nd PSU, it just doesn't work right. Slowly got to 475 MHZ, but not consistent. Going to try putting back the 3 boxes + 1 controller board back on the first PSU, and then adding the 2 on the 2nd psu and see if I can get anywhere

[markm]

I switched the failover sequence of my merged minings pools to put bitparking's mmpool first instead of my own p2pool, and boom, I am now seeing 3.35 terrahashes on the LCD readout. So somehow p2pool was causing me to lose 2/3 of my hashing power.

I tried removing geistgeld, the fastest block time merged mined coin, such fast blocks that mmpool does not merge them, from my merge but still as soon as I switched back to using the p2pool pool the hash rate shown on the LCD started dropping back down again.

So I guess it must be something about my p2pool, rather than the fact that I merge more coins with it than mmpool does, that is causing such massive losses.

-MarkM-

[bitpop]

I switched the failover sequence of my merged minings pools to put bitparking's mmpool first instead of my own p2pool, and boom, I am now seeing 3.35 terrahashes on the LCD readout. So somehow p2pool was causing me to lose 2/3 of my hashing power.

I tried removing geistgeld, the fastest block time merged mined coin, such fast blocks that mmpool does not merge them, from my merge but still as soon as I switched back to using the p2pool pool the hash rate shown on the LCD started dropping back down again.

So I guess it must be something about my p2pool, rather than the fact that I merge more coins with it than mmpool does, that is causing such massive losses.

-MarkM-

P2pool does require extreme cpu for a  few seconds when generating work every 30 seconds

Back in the day we were seeing 30 second delays with Bitcoin getting the block ready to mine

[markm]

GeistGeld's block target is something like 10 or 12 seconds I think, not 30 seconds.

But still, even leaving GeistGeld out of the merge I was losing close to 2/3 of my hashing power it seemed.

With only my Saturn and lots of Klondikes and Block Eruptors my losses if any were much lower. It seems the more hashing you throw at p2pool the higher the loss rate becomes.

Maybe I need to use solid state disks for my coin daemons to speed up block creation?

Though I think all transactions should be in RAM so maybe that would not really make the difference.

-MarkM-

[bitpop]

Idk but mmpool is great

[Elenelen]

if anyone is interested, I have plates that replace the mounting apparatus they used.  It will get the foam out of the way of the VRMs and allow more airflow over the PCB.

They'll come with the plates, screws, nuts and spacers for all 5 ASICs - .0675 BTC plus shipping (USPS Priority/Priority Express, your choice)
PM me ...
Ski

Sorry but they will not fit !.... you need an T-shape plate, and not an I-shape

true, it won't fit with the box on it.  You'd have to turn the Arctic radiator 90^o and it will be mounted in the same fashion as the Jupiter.  It looks like everyone is doing their best to make it work. 
Just sayin' ...

Then your 90 degrees turned I-bar will overlap (or even touch) the VRM's, which is just bad....  You may do "your best" to make things work, but you haven't tested it and you sell your bars for $$

[jelin1984]

also what difficulty per neptne at pool

1024 or more?

[AFox]

OK, I have a modified-DIY-box below 70 C and running on 500 Mhz:

How do I get above 500 Mhz ?

Hi,

This may be something our engineers implement in future firmware revisions, but this is the max limit at the moment I am afraid.

Kind regards,
Med vänlig hälsning |  Best regards
Kurt Radcliffe
Kncminer

[bitpop]

I wonder what happens if you firebug the html form