Swedish ASIC miner company kncminer.com

[Aggrophobia]

I purchased 2 of the recent fire-sale modules for my 4-port Jupiter, but needed to add a couple of pin headers to support all six modules.

I've never really soldered anything before and had no idea what I was doing, but I managed to get everything working! It took a few tries though. I made the assumption that the two blank spaces were simply solder filled holes, but no matter how hot I heated them, I couldn't melt them. I ended up just soldering to the surface.

I'm currently running my existing 4 modules on ports 2-5 so that I can simply attach the new ones onto the side of my open-case Jupiter using ports 1 and 6.

The point of me sharing this is simply to show my excitement and proof that even a noob like me can make this work!

OMFG...are you serious??? 

[1714975329]

1714975329

[1714975329]

1714975329

[edgar]

So Awesome r1senfa17h!!

i would be so scared of voiding the warranty, especially with no modules to add.....however, you did great!

If only KnC had done the job fairly for all rigs 'we' wouldnt need to risk voiding warranty/destroying our rigs AND those '200 only' modules might have sold out in 30 seconds instead of 3 minutes....

BiG IFs!

[Bogart]

So Awesome r1senfa17h!!

i would be so scared of voiding the warranty, especially with no modules to add.....however, you did great!

If only KnC had done the job fairly for all rigs 'we' wouldnt need to risk voiding warranty/destroying our rigs AND those '200 only' modules might have sold out in 30 seconds instead of 3 minutes....

BiG IFs!

They also may have been delivered later, having taken the extra time to do that.

There is also still the issue of not having space in the case for 6 modules.

I don't know if I'd call the job "great", but he had success, so that still counts.

[1l1l11ll1l]

Here's my all-in-one Jupiter+2

[xyzzy099]

Here's my all-in-one Jupiter+2

<Images edited out>

Wow, well done

[Paladin69]

Here's my all-in-one Jupiter+2

Dude, that is bad ass.  What coolers did you buy?

EDIT:  Also, well done r1senfa17h

[DimensionsOfHell]

I just got my board about an hour ago. Freaking UPS took all day to deliver the damn thing. It said "Out for delivery" at 5:45AM, and it just got delivered to me at 4:58PM. They took freakin 11 hours just to do a local delivery. Anyway...

I added in the [1] upgrade board I ordered. It sits in the front right row, as my Saturn came with the back populated instead of the front. My Saturn also did not come with the air divider that the Jupiter's have, so I had to cut up a piece of cardboard box and divide the air that is coming out of the front board. The rear board was starting to get to 80c.

Now my boards are:

Mining Status

ASIC slot #1   70.5 ℃ (This is the board behind the front board)
ASIC slot #2   -
ASIC slot #3   69.5 ℃
ASIC slot #4   66.5 ℃
ASIC slot #5   -
ASIC slot #6   -

Also quick note for other who are getting their upgrade boards: KnC did NOT include screws for the boards to mount to the case, HOWEVER, look in your Arctic cooler box, there should be a bag of screws. The smallest screws there [should be 4 of them] will fit to secure the board to the case. Just use two of them.

KnC also did NOT include screws to hold the extra aluminum plate, that they include in your upgrade kit, that hold the plate in place on the cooler. You can use the remaining 2 small screws from the Arctic cooler, and screw in from the bottom up. It is very small, but it is enough to ensure that the plate does not move.


On just the two boards, I was averaging 245GH/s (I sent one board in for RMA that was only hashing at half speed, and they sent it back so that it hashed at 3/4 speed) on FW 0.99. After I added in the 3rd board it was hashing at a total of 355GH/s (20 min average).

I downgraded the FW to 0.9.8.1(beta), and the rate shot up quickly to 370GH/s within 2-3 minutes. So far after 40 minutes (as of this writing) on FW 0.9.8.1(beta), I am currently averaging 410Gh/s. So it seems the other die that was not working properly might finally be working.

If the bad die still wasn't fully working, I would've only been able to hash at a maximum of 389GH/s (245 (previous average) + 144 (best performance expectation)).

So for any of you guys that might still have bad dies, and are adding a new module(s);

Add in new module in FW 0.99 and let it run for a few mins to see how it goes, if things aren't better, try out FW 0.9.8.1(beta)

Hope this helps someone!

I'm very happy that my unit is finally working close to as it should be! :-)


EDIT: This is off KnC topic but I've gotta ask (and I didn't want to add another off topic post) because I'm confused with what is going with the Eligius pool.



Those last two payments, how come they say "S" on them? I've received the 0.1995... BTC payment on the top , but not the 0.1757... BTC payment that was supposed to be on the 17th. I don't want to mine on Eligius if I'm not going to be getting my BTC payments. Anyone know if that is normal? Some payments get delayed a long time?

Sorry for the off topic questions, but I would really appreciate an answer from someone that knows what is going on. I don't want to lose $128 (current exchange rate).

[Phoenix1969]

Yep, I went back to beta as well. 0.99 just isn't as fast for me either.
the last 3 hours look to be alot less at the pool..
http://eligius.st/~wizkid057/newstats/userstats.php/1LnqQzhbp1udcqByNZWnveHXJorDuJHWRs
just switched back 5 mins ago...

[ncs0ne]

Bitcoinorama mentioned "ghosts in the system", remember ?
Now I had a visit.
I changed the hosting location to another room. I guess they don't like this ?

I shut down miners via "shutdown -h now", then PSU off, after the light turns off. I'm used to do it this way, feels better than hard kill through PSU switch off, what corrupted my system image once ago. Simple Reset did not help, had to disconnect all boards and stuff until it finally worked again.

Ok, new location, plug in, start up. The light and everything blinking ok.
SSH to miner1 takes longer than usual, but successful. Same for #2.

Then cgminer had no connection to the pool on both miners.
cmd "route" output looked good, ping from miner 1->2 & 2-> 1 was fine, ping from miner1&2->laptop was fine, ping from laptop to router was fine, ping from miner1+2 -> router TIMEOUT! So confusing  

As Next, I shutdown & restarted the network switch, did not help. Then switching ports, cables, plug off/on the power-line ethernet and all other network hardware involved. Miners shutdown as well. Restarted step by step all hardware, tried stuff on its own and as last the miners again.

Nothing changed. Still takes ages for ssh to connect, but successful.
Setting network config in the webgui did not help either. Looking onto my router none of both has requested an IP address nor assigned, my router told me. All the time I connected via wifi to this router and through it to the miners, to their fixed IPs they get from the router.   !!

Scratching my head, I thought ok let's see what "ps | grep dhc" shows, then I killed the dhcp-client and restarted it with same parameters shown by the mentioned ps cmd.

Code:

udhcpc -b -x hostname MyMinersHostname eth0

BINGO! That helped to scare the ghosts away, router showed me immediately registered clients (both miners), ssh was working as fast as usual again and they were already hashing away when I was checking them. ~30min downtime hurts, I need to learn some digital voodoo or else to prevent similar stuff when changing the location next time  

[soy]

New module in and my Merc is now a Saturn.  Interestingly, the new module is running 11°C cooler than the original.  Stats look good.
KnC 0: | 287.3G/286.4Gh/s | A:28534 R:11 HW:513 WU:3915.7/m

All cores good.

I must say since I'm use to looking at cgminer with a number of USB BE's, each with its own descriptive line, when I saw only KNC0 I started troubleshooting - until I realized it was hashing at 280 not 140.

[xyzzy099]

New module in and my Merc is now a Saturn.  Interestingly, the new module is running 11°C cooler than the original.  Stats look good.
KnC 0: | 287.3G/286.4Gh/s | A:28534 R:11 HW:513 WU:3915.7/m

All cores good.

I must say since I'm use to looking at cgminer with a number of USB BE's, each with its own descriptive line, when I saw only KNC0 I started troubleshooting - until I realized it was hashing at 280 not 140.

FWIW, if you install Bertmod 0.3 it gives you the option of using BFGMiner, and it will show each module as a separate device (KNC0, KNC1, etc.) and even shows exactly which core returns each share.  I don't think it will disable cores that are generating h/w errors though.

[1l1l11ll1l]

Here's my all-in-one Jupiter+2


Images Removed

Dude, that is bad ass.  What coolers did you buy?

EDIT:  Also, well done r1senfa17h

The Corsair H60. They are both the H60, however there's 2 different models. The model with the ribbed hose, the other one's pump melted, the bearings went out and the bloody pump melted.

Then I just fastened the included fans to the radiators through the case, this gives a nice solid mount, and moves the heat out okay.
I did have to remove the angled aluminum fins that are (were) mounted to the lid.

The placement is also ideal, as the VRMs get plenty of airflow. Please note I'm using substantially upgraded front fans (D14SH-12).

The pumps gain their power from the fan connector on the module, while the front and rear fans are connected to Molex power.

[CYPER]

Nov 15th update:

The Bad news is the delivery of parts which we needed to begin today has been delayed by about 5-10 days

There' no issue with parts, just the guys developing the final revision (hardware, and software) wanted to optimize it further, so it's more ongoing performance improvements.

Which is it?  Did I miss a subsequent clarification?

I was told the same thing while in Stockholm - that there is no issue with parts. I guess it's just miscomunication as usual.

The point of me sharing this is simply to show my excitement and proof that even a noob like me can make this work!

Officially, I cannot condone what you have done! you have voided your warranty.

(Unoffically; Result! )

Oficially we are not supposed to even open the case or remove modules ourselves, but we do, because we have to and KNC approves it, so this warranty business is like a gate in a field  

Are these controller boards available from KnC in case of a failure? Will the new controller boards be compatible with the current upgrade modules, or vice versa?

No, they don't sell them or at least that's what I was told.

[opentoe]

Besides KNC forgetting the screws both my modules are running hot. Did anyone put thermal grease in between the aluminum plate and the heatsink or just greased the chip only? Both new modules I put in are running 10 degrees warmer then the other two. Seriously, how warm can these go before there starts to be a problem? I have dies shutting down all over the place on both new modules. I'm only hashing at 540Gh/s and according to BTC Guild, it's only 318Gh/s. Although BTC Guild may need some time to catch up, since it's only been a 10 minutes. Yea, I posted this on the KNC forums too...I want to cover all bases here.

[edgar]

i still think r1sen did a great job, regardless!

but 1l1l11ll1ls effort is 5 star!!

That is a WIN!! i have to applaud it!!

i wonder who will have the balls to poopoo this creation...


BTW - adding two 'extra' connectors during production would hardly have slowed down SEPTEMBER October rigs! dont be so daft!!

by the same logic, the inclusion of module mounting screws would have slowed things down too much as well! (or cost too much?)
 
Again, Daft!!



STILL interested to hear the KnC official stance on PCI-express connectors on the v850s

are you hosting them with one 'twin' cable heating up like mine did for a month before dying or are you in possession of extra cables?

also if the heat generated is a cause or an effect of the 0 Die low current issue?

very interested to hear a response as my v850 up & died yesterday...

[r1senfa17h]

Well, I've got 5 modules running now and I'm getting about 660GH/s. Adding only 110GH/s to my old 550GH/s seemed slow, so I used bertmod and see that I've got the dreaded Die 0 issue! I'm running firmware 0.98.1 and blocked the airflow to raise the temp without success (reached 80 degrees C before I lost my nerve and started the fan again).

Are there any other tricks before I decide to RMA it? My 2nd module arrives tomorrow and I'm hoping for better luck.

[soy]

Here's my all-in-one Jupiter+2

Very professional.

[soy]

I just got my board about an hour ago. Freaking UPS took all day to deliver the damn thing. It said "Out for delivery" at 5:45AM, and it just got delivered to me at 4:58PM. They took freakin 11 hours just to do a local delivery. Anyway...

I added in the [1] upgrade board I ordered. It sits in the front right row, as my Saturn came with the back populated instead of the front. My Saturn also did not come with the air divider that the Jupiter's have, so I had to cut up a piece of cardboard box and divide the air that is coming out of the front board. The rear board was starting to get to 80c.

Now my boards are:

Mining Status

ASIC slot #1   70.5 ℃ (This is the board behind the front board)
ASIC slot #2   -
ASIC slot #3   69.5 ℃
ASIC slot #4   66.5 ℃
ASIC slot #5   -
ASIC slot #6   -

Also quick note for other who are getting their upgrade boards: KnC did NOT include screws for the boards to mount to the case, HOWEVER, look in your Arctic cooler box, there should be a bag of screws. The smallest screws there [should be 4 of them] will fit to secure the board to the case. Just use two of them.

KnC also did NOT include screws to hold the extra aluminum plate, that they include in your upgrade kit, that hold the plate in place on the cooler. You can use the remaining 2 small screws from the Arctic cooler, and screw in from the bottom up. It is very small, but it is enough to ensure that the plate does not move.


On just the two boards, I was averaging 245GH/s (I sent one board in for RMA that was only hashing at half speed, and they sent it back so that it hashed at 3/4 speed) on FW 0.99. After I added in the 3rd board it was hashing at a total of 355GH/s (20 min average).

I downgraded the FW to 0.9.8.1(beta), and the rate shot up quickly to 370GH/s within 2-3 minutes. So far after 40 minutes (as of this writing) on FW 0.9.8.1(beta), I am currently averaging 410Gh/s. So it seems the other die that was not working properly might finally be working.

If the bad die still wasn't fully working, I would've only been able to hash at a maximum of 389GH/s (245 (previous average) + 144 (best performance expectation)).

So for any of you guys that might still have bad dies, and are adding a new module(s);

Add in new module in FW 0.99 and let it run for a few mins to see how it goes, if things aren't better, try out FW 0.9.8.1(beta)

Hope this helps someone!

I'm very happy that my unit is finally working close to as it should be! :-)


EDIT: This is off KnC topic but I've gotta ask (and I didn't want to add another off topic post) because I'm confused with what is going with the Eligius pool.



Those last two payments, how come they say "S" on them? I've received the 0.1995... BTC payment on the top , but not the 0.1757... BTC payment that was supposed to be on the 17th. I don't want to mine on Eligius if I'm not going to be getting my BTC payments. Anyone know if that is normal? Some payments get delayed a long time?

Sorry for the off topic questions, but I would really appreciate an answer from someone that knows what is going on. I don't want to lose $128 (current exchange rate).

I got the module and an i30.  I didn't fasten the bar to the permanent piece on the i30 - probably should have.  Did remove the plastic.  Did take the time to position the cooling tower as exactly centered, fore and aft, on the module while side to side I judged by the relative position of the permanent bar on the i30 and the clamp down bar.  Interestingly looking at the exposed corners of the ASIC module there seems to be more exposed under the original (an A30?).  I don't think there's more copper on the i30 than the A30, is there?  Probably just these aging eyes.

[soy]

I just got my board about an hour ago. Freaking UPS took all day to deliver the damn thing. It said "Out for delivery" at 5:45AM, and it just got delivered to me at 4:58PM. They took freakin 11 hours just to do a local delivery. Anyway...

I added in the [1] upgrade board I ordered. It sits in the front right row, as my Saturn came with the back populated instead of the front. My Saturn also did not come with the air divider that the Jupiter's have, so I had to cut up a piece of cardboard box and divide the air that is coming out of the front board. The rear board was starting to get to 80c.

Now my boards are:

Mining Status

ASIC slot #1   70.5 ℃ (This is the board behind the front board)
ASIC slot #2   -
ASIC slot #3   69.5 ℃
ASIC slot #4   66.5 ℃
ASIC slot #5   -
ASIC slot #6   -

Also quick note for other who are getting their upgrade boards: KnC did NOT include screws for the boards to mount to the case, HOWEVER, look in your Arctic cooler box, there should be a bag of screws. The smallest screws there [should be 4 of them] will fit to secure the board to the case. Just use two of them.

KnC also did NOT include screws to hold the extra aluminum plate, that they include in your upgrade kit, that hold the plate in place on the cooler. You can use the remaining 2 small screws from the Arctic cooler, and screw in from the bottom up. It is very small, but it is enough to ensure that the plate does not move.


On just the two boards, I was averaging 245GH/s (I sent one board in for RMA that was only hashing at half speed, and they sent it back so that it hashed at 3/4 speed) on FW 0.99. After I added in the 3rd board it was hashing at a total of 355GH/s (20 min average).

I downgraded the FW to 0.9.8.1(beta), and the rate shot up quickly to 370GH/s within 2-3 minutes. So far after 40 minutes (as of this writing) on FW 0.9.8.1(beta), I am currently averaging 410Gh/s. So it seems the other die that was not working properly might finally be working.

If the bad die still wasn't fully working, I would've only been able to hash at a maximum of 389GH/s (245 (previous average) + 144 (best performance expectation)).

So for any of you guys that might still have bad dies, and are adding a new module(s);

Add in new module in FW 0.99 and let it run for a few mins to see how it goes, if things aren't better, try out FW 0.9.8.1(beta)

Hope this helps someone!

I'm very happy that my unit is finally working close to as it should be! :-)


EDIT: This is off KnC topic but I've gotta ask (and I didn't want to add another off topic post) because I'm confused with what is going with the Eligius pool.



Those last two payments, how come they say "S" on them? I've received the 0.1995... BTC payment on the top , but not the 0.1757... BTC payment that was supposed to be on the 17th. I don't want to mine on Eligius if I'm not going to be getting my BTC payments. Anyone know if that is normal? Some payments get delayed a long time?

Sorry for the off topic questions, but I would really appreciate an answer from someone that knows what is going on. I don't want to lose $128 (current exchange rate).

G is auto payout S is manual payout, or vice versa.

[soy]

Bitcoinorama mentioned "ghosts in the system", remember ?
Now I had a visit.
I changed the hosting location to another room. I guess they don't like this ?

I shut down miners via "shutdown -h now", then PSU off, after the light turns off. I'm used to do it this way, feels better than hard kill through PSU switch off, what corrupted my system image once ago. Simple Reset did not help, had to disconnect all boards and stuff until it finally worked again.

Ok, new location, plug in, start up. The light and everything blinking ok.
SSH to miner1 takes longer than usual, but successful. Same for #2.

Then cgminer had no connection to the pool on both miners.
cmd "route" output looked good, ping from miner 1->2 & 2-> 1 was fine, ping from miner1&2->laptop was fine, ping from laptop to router was fine, ping from miner1+2 -> router TIMEOUT! So confusing  

As Next, I shutdown & restarted the network switch, did not help. Then switching ports, cables, plug off/on the power-line ethernet and all other network hardware involved. Miners shutdown as well. Restarted step by step all hardware, tried stuff on its own and as last the miners again.

Nothing changed. Still takes ages for ssh to connect, but successful.
Setting network config in the webgui did not help either. Looking onto my router none of both has requested an IP address nor assigned, my router told me. All the time I connected via wifi to this router and through it to the miners, to their fixed IPs they get from the router.   !!

Scratching my head, I thought ok let's see what "ps | grep dhc" shows, then I killed the dhcp-client and restarted it with same parameters shown by the mentioned ps cmd.

Code:

udhcpc -b -x hostname MyMinersHostname eth0

BINGO! That helped to scare the ghosts away, router showed me immediately registered clients (both miners), ssh was working as fast as usual again and they were already hashing away when I was checking them. ~30min downtime hurts, I need to learn some digital voodoo or else to prevent similar stuff when changing the location next time  

While running the Merc with a single module, at times a restart would hang, cat5 RJ54 port lights flashing but no white light and no red/green.  I would shut down, move the ribbon cable to another connector on the controller board, reboot, and Bob's my uncle.