OK... one picture with the 3 boxes:
Are those riveted or screwed together? Any chance of popping open a case and taking a pic?
Topic: Swedish ASIC miner company kncminer.com - page 373. (Read 3050076 times)
Are those riveted or screwed together? Any chance of popping open a case and taking a pic?
OK... one picture with the 3 boxes:
You can see the display (bleu thing on the bottom of the Controller board)
You can see the display (bleu thing on the bottom of the Controller board)
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working 
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
Generally speaking that means the core those are tied to is not coming online.. Do you have the option to raise the voltage? If so the kick procedure is to basically up the voltage temporarily to get it to turn on, if it starts hashing you can then turn it back down..
A power down, and new restart did the work... all cores are OK now.
But I noticed that, a few times some of the boxes stop Hashing completely (others keep running).... then a restart CGminer brings them back again.... I think there are still some bugs in the driver software...
But no support today >> looks like KnC is on Swedish holiday.
The LEPA 1600W PSU is close to capacity, as others said if it keeps happening definitely try to put one board on another PSU to see what happens..
In addition the LEPA 1600W only exposes 4 of it's 6 12V rails via the PCIe connectors so even with separate cables at least two of your boxes is sharing a rail unless you build a custom cable to pull from 12V1 or 12V2 vice 3-6, which are all you touch across the 6 red PCIe connectors..
Yes, I know: that's why I have two Lepa's
My mistake, apparently I struggle to read sometimes..
3.3 th with how many boxes
5 boxes or 3 boxes?
5 boxes or 3 boxes?
There's quite some Tuning needed (which is understandable with the first FW).
The 3-box system I tested first was much more stable then the 5-box system. I don't get the expected 3.3Th (which could be calculated from the 3-boxes) but it's only 3.0Th (and the 5s average is jumping all over the place, which was steady in the 3 box setup).
It's just like the first Jupiters..... mine ran only 450Ghs in the first week, then after several FW upgrades it was 550 (and with OC 650).
The 3-box system I tested first was much more stable then the 5-box system. I don't get the expected 3.3Th (which could be calculated from the 3-boxes) but it's only 3.0Th (and the 5s average is jumping all over the place, which was steady in the 3 box setup).
It's just like the first Jupiters..... mine ran only 450Ghs in the first week, then after several FW upgrades it was 550 (and with OC 650).
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
Generally speaking that means the core those are tied to is not coming online.. Do you have the option to raise the voltage? If so the kick procedure is to basically up the voltage temporarily to get it to turn on, if it starts hashing you can then turn it back down..
A power down, and new restart did the work... all cores are OK now.
But I noticed that, a few times some of the boxes stop Hashing completely (others keep running).... then a restart CGminer brings them back again.... I think there are still some bugs in the driver software...
But no support today >> looks like KnC is on Swedish holiday.
The LEPA 1600W PSU is close to capacity, as others said if it keeps happening definitely try to put one board on another PSU to see what happens..
In addition the LEPA 1600W only exposes 4 of it's 6 12V rails via the PCIe connectors so even with separate cables at least two of your boxes is sharing a rail unless you build a custom cable to pull from 12V1 or 12V2 vice 3-6, which are all you touch across the 6 red PCIe connectors..
Yes, I know: that's why I have two Lepa's
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
Generally speaking that means the core those are tied to is not coming online.. Do you have the option to raise the voltage? If so the kick procedure is to basically up the voltage temporarily to get it to turn on, if it starts hashing you can then turn it back down..
A power down, and new restart did the work... all cores are OK now.
But I noticed that, a few times some of the boxes stop Hashing completely (others keep running).... then a restart CGminer brings them back again.... I think there are still some bugs in the driver software...
But no support today >> looks like KnC is on Swedish holiday.
The LEPA 1600W PSU is close to capacity, as others said if it keeps happening definitely try to put one board on another PSU to see what happens..
In addition the LEPA 1600W only exposes 4 of it's 6 12V rails via the PCIe connectors so even with separate cables at least two of your boxes is sharing a rail unless you build a custom cable to pull from 12V1 or 12V2 vice 3-6, which are all you touch across the 6 red PCIe connectors..
>> looks like KnC is on Swedish holiday.
They've been on holiday for months now. Just wait until August when they get drunk out of their fucking minds and spend all night vomiting in the streets of Stockholm. We won't hear a word out of them then. I'll be getting my Neptune in late September - if I am lucky.
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
did you say you're using 1600W LEPA PSU, was that 1 or 2 of them?
if you're using 1 only, then you're Neptune is probably starving for more power.
No two Lepa's are now connected (one does 3 boxes the other 2 boxes plus the controller board).
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
Generally speaking that means the core those are tied to is not coming online.. Do you have the option to raise the voltage? If so the kick procedure is to basically up the voltage temporarily to get it to turn on, if it starts hashing you can then turn it back down..
A power down, and new restart did the work... all cores are OK now.
But I noticed that, a few times some of the boxes stop Hashing completely (others keep running).... then a restart CGminer brings them back again.... I think there are still some bugs in the driver software...
But no support today >> looks like KnC is on Swedish holiday.
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
did you say you're using 1600W LEPA PSU, was that 1 or 2 of them?
if you're using 1 only, then you're Neptune is probably starving for more power.
Easier and better = make video and upload to youtube.
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
Generally speaking that means the core those are tied to is not coming online.. Do you have the option to raise the voltage? If so the kick procedure is to basically up the voltage temporarily to get it to turn on, if it starts hashing you can then turn it back down..
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
Oh fuck. This is ugly. Are you serious? The Neptune doesn't work? They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
Here it goes again...
Cardboard, Hair dryers, all the tricks from the Jupiters.
I am already internally laughing at anyone who ordered a non-refundable titan from them....
Tbh the Titan is more likely to turn a profit than the neptunes, however that's not saying much is it ;P
I now have 5 boxes running, but 3 boxes have 2 DC/DC's not working
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
They show: DC0 = 0.0035V 0.0000A 0.0000W 25C
Does anyone know the procedure to kick them ?
The one downside of canceling my Neptune is I do not have a resource to hack on and make a firmware like I did (currently 1.01.1-padrino) for the current systems.. Running November boards at 250Gh/s (250W) with about 1.5Th/s at the pool for a 6 prot, wonder how far the Neptune ASIC can be pushed..
It's a shame but given how hard the single PCI-E connectors are being pushed already I don't think there will be much room for overclocking.
Indeed: the "Advanced window" has a Mhz select-list: from 50 to 500 Mhz..... but default is 475 already.
The current systems are not much different, the standard firmware only allows 1000Mhz for the November systems, 900Mhz is stock... I capped my custom firmware at somewhere around 1600Mhz and with a quick cooling mod most run that way, hoping for similar results with these once it can be hacked on..
The one downside of canceling my Neptune is I do not have a resource to hack on and make a firmware like I did (currently 1.01.1-padrino) for the current systems.. Running November boards at 250Gh/s (250W) with about 1.5Th/s at the pool for a 6 prot, wonder how far the Neptune ASIC can be pushed..
It's a shame but given how hard the single PCI-E connectors are being pushed already I don't think there will be much room for overclocking.
I was thinking the same but if they are like the old boards they actually have space for a second connector on the PCB, I currently mod VRMs and other goodies on the boards, dropping in a second PCIe connector will definitely be needed..
I push 250W via one connector on the current boards and have one on my desk that melted the connector and the cable together.. If the new ones run 300W stock it's crazy, I would however lean toward LEPA PSUs for trying anything as they run thicker wires for their PCIe cables and try and drop in a second molex..
If they designed things similar to the current chips we could potentially see 4.5-5Th/s with a Neptune however with the VRM configuration on the Neptune boards I am wondering how much headroom is left, there simply may not be too much to work with on these boards..
Would love to inspect a board and look at the stock voltage/amps being pulled..
The one downside of canceling my Neptune is I do not have a resource to hack on and make a firmware like I did (currently 1.01.1-padrino) for the current systems.. Running November boards at 250Gh/s (250W) with about 1.5Th/s at the pool for a 6 prot, wonder how far the Neptune ASIC can be pushed..
It's a shame but given how hard the single PCI-E connectors are being pushed already I don't think there will be much room for overclocking.
Indeed: the "Advanced window" has a Mhz select-list: from 50 to 500 Mhz..... but default is 475 already.
Jump to: