Topic: KnCMiner Jupiter Miner First Impressions - page 17. (Read 65993 times)

LAST CORE HAVE 48 OFF
AND 15A

ALSO AND MY OTHER CORE HAVE NOT A LOT OF AMBER

I HAVE BOARDS WITH 8 VRMs. ALL

MINE WITH 0.91 FIRMWARE

ASIC Board    Info
0    

Temperature sensor: 44.5 C
Die ID    Cores ON    Cores OFF    %
0   48   0    100
1   48   0    100
2   48   0    100
3   48   0    100
DC/DC ID    ON/OFF    Status    Input Voltage    Output Voltage    Output Current
0    ON    OK    11.9 V    0.71 V    39.4 A (28 W)
1    OFF    OFF    12 V    0.706 V    0 A
2    ON    OK    11.8 V    0.753 V    40.1 A (30.2 W)
3    OFF    OFF    11.9 V    0.736 V    0 A
4    ON    OK    12 V    0.735 V    38.1 A (28 W)
5    OFF    OFF    11.9 V    0.703 V    0 A
6    OFF    OFF    12 V    0.741 V    0 A
7    ON    OK    12 V    0.755 V    40.6 A (30.7 W)
1    

Temperature sensor: 56.0 C
Die ID    Cores ON    Cores OFF    %
0   48   0    100
1   48   0    100
2   48   0    100
3   48   0    100
DC/DC ID    ON/OFF    Status    Input Voltage    Output Voltage    Output Current
0    ON    OK    11.9 V    0.735 V    39.7 A (29.2 W)
1    OFF    OFF    11.9 V    0.712 V    0 A
2    ON    OK    12 V    0.711 V    41.4 A (29.4 W)
3    OFF    OFF    11.9 V    0.741 V    0 A
4    ON    OK    12 V    0.722 V    40.8 A (29.5 W)
5    OFF    OFF    12 V    0.71 V    0 A
6    OFF    OFF    12 V    0.729 V    0 A
7    ON    OK    12 V    0.735 V    39.7 A (29.2 W)
2    

Temperature sensor: 46.0 C
Die ID    Cores ON    Cores OFF    %
0   45   3    93.8
1   48   0    100
2   48   0    100
3   48   0    100
DC/DC ID    ON/OFF    Status    Input Voltage    Output Voltage    Output Current
0    ON    OK    11.8 V    0.742 V    37.6 A (27.9 W)
1    OFF    OFF    11.8 V    0.713 V    0 A
2    ON    OK    11.8 V    0.727 V    40 A (29.1 W)
3    OFF    OFF    11.8 V    0.739 V    0 A
4    ON    OK    11.8 V    0.75 V    39.2 A (29.4 W)
5    OFF    OFF    11.8 V    0.708 V    0 A
6    OFF    OFF    11.8 V    0.729 V    0 A
7    ON    OK    11.8 V    0.749 V    39.9 A (29.9 W)
3    

Temperature sensor: 51.0 C
Die ID    Cores ON    Cores OFF    %
0   48   0    100
1   48   0    100
2   48   0    100
3   0   48    0
DC/DC ID    ON/OFF    Status    Input Voltage    Output Voltage    Output Current
0    ON    OK    11.9 V    0.703 V    38.9 A (27.3 W)
1    OFF    OFF    11.9 V    0.697 V    0 A
2    ON    OK    12 V    0.734 V    41.2 A (30.2 W)
3    OFF    OFF    11.9 V    0.733 V    0 A
4    ON    OK    12 V    0.695 V    39.7 A (27.6 W)
5    OFF    OFF    12 V    0.691 V    0 A
6    OFF    OFF    11.9 V    0.819 V    0 A
7    ON    OK    12 V    0.818 V    15.9 A (13 W)

the other miner that has one board with 8 VRMs does run 10C cooler than boards with 4 VRMs.  That miner is not in question. I leave that one alone

I doubt my other miner has a bad PSU since it has done well even with all the miner's troubles

Which is the modded firmware
And where I can find that!?

the engineer in me made me refuse to quit on this miner


The 5 times button will bring you back to overclocking if you upgrade to pre .95 after doing it.  If you have .94, .93 etc and use the button it won't switch until flash it after the reset.


Since I have one bad VRM I can't run .93, .92 or .91 but I can run .94 on it without the FPGA error loop

This allowed me to bring all 4 boards overclocking again but with .94 not .93.

I have gained substantial Gh (avg 430-440Gh now, but have seen it spike up a few times when it tries to bring back cores. cgminer hasn't restarted yet on it after 30 mins)

I do have a permanent 48 cores dead due to the bad VRM now.
 (the modded firmware can't read that they are off when VRM is in this state)

Here are the stats from about 30 minutes of hashing:

avg Gh from BTCGuild  ~445  (75 Gh increase from running .95, it was ~520 before the 'upgrade' to .95)


ASIC slot #1: 46.5 ℃   (the bad VRM board - back left)
ASIC slot #2: -
ASIC slot #3: 50.0 ℃   (front left)
ASIC slot #4: 44.5 ℃   (front right)
ASIC slot #5: 55.5 ℃   (back right)
ASIC slot #6: -


Total DC/DC power output: 683 W    (used to run 770watts when ~520Gh)
















WU steady 7500-7600 which is about 1000 more than when it ran .95 but about 600 less than before messing with .95

other miner on .93 has been running WU over 8200 for days

I did do the 5 times/wait 5 seconds/5 times button two times over the course of the 8 hours or so testing



I do that but how I understand that I had do hardware reset to my miner!?
Any lights or something?




How overckloded the bad board?


Ddpos the Jupiter with 8 VRm how is the amper at all board?
Because at mine at all board is only 30 a non one at 50a
Maybe is psu faulty?
I have corsair ax 850

I did do the 5 times/wait 5 seconds/5 times button two times over the course of the 8 hours or so testing

doesn't seem to help except it might of allowed the overclocking to switchback when using .93 (I forget if that was how I got it go back a few times)


If I hit the miner status page as soon as it first starts I have seen some amps come from that bad VRM briefly (caught it showing 17 amps once) but it always quickly goes to under 1 amp everytime

I have no pattern for the two boards that take turns having one die that avgs 50% off cores.  It changes from different dies on the ASIC as well so it is some type of limiting situation where it can't have all cores running with more than two active boards in the miner anymore.

when I tested just those two boards and able to get it to switchback to overclock voltage levels, they didn't show that 50% bad cores. It is not simple to get it to go back to overclocking. Perhaps the factory resets that I did do allowed that to happen. But then I couldn't run 3 boards overclocked and good anyway so the Gh is the same as all 4 running .95 now  (the most I could hope for 3 boards would be 140Gh each for 420 but that never happened.. but I did get 280 with two boards..)

sad situation since my other miner is a ticking timebomb

I may try a few more times with that blown miner and hope for luck

 

Have you attempted to reset your KnCs control board?

the PSU on that one has been a hero, was running stable 770watts even though it is just a 850watter when it was 520Gh on .93

http://www.microcenter.com/product/410331/Hale90_Series_V2_850_Watt_Modular_Power_Supply


fact is, the VRMs got cooked with firmware not made for 4 VRM boards - changing firmware after the fact degraded it and made it unstable

also, when shutting down the hundred times last night testing I always quit cgminer, then ran 'shutdown -h now'
then I would unplug the pin connection (I have a premade adapter plug) and then powerswitch the PSU.

PSU type??

be glad if you have 8 VRMs.  be very glad.  Here is my sad experience with .95 with one of my miners from last night (till 5am this morning)


After going to .95 on my miner with all boards 4 VRMs, it blew a VRM.  The blown one still gets voltage but outputs < 1 amp current.  Also, two other boards struggle with lots of cores deactivating resulting in ~370 Gh for the miner on .95

What has happened of course, was that since I was forced to overclock a 4 VRM miner with firmware not designed for it (pre .95) it is now unstable at the lower voltage/current/wattage.

If I disconnect the board with the blown VRM and just plug in the two boards with inconsistent cores and run .93, they hash very well IF the VRMs switchback to run at the higher overclock levels, but sometimes going back to pre .95 firmwares will not raise the voltage and you won't get any advantage)  Around ~270 Gh like a good Saturn if they switchback and overclock.  But if I add the third good board and run either .93 or .94 I see cores turning off again from those two boards.  The blown VRM board can not run .93, .92 or .91 anymore and goes into the FPGA endless error loops.

I am stuck with lessor of all evils of running all boards with .95 since that at least drops the wattage and allows the blown VRM board to still hash somewhat (saw it peak at 100 Gh solo for a short time on .95 so I know it can still output with 3 VRMs functioning)

So it costs me 150Gh to be the testbed for KNC. And of course my other miner with the one 8 VRM board is still running .93 and 520-530 Gh stable.  If it ever changes firmware it will no doubt run poorly too because of the three baked 4VRM boards.

.95 overnight



ASIC slot #1: 54.0 ℃
ASIC slot #2: 54.5 ℃
ASIC slot #3: 58.0 ℃
ASIC slot #4: 52.5 ℃

cover on, ambiente ~22 ℃

Here is the results after 13 hours on firmware v0.9.5.

I have also included the screenshot of the previous 13 hour result on firmware v0.9.4 to compare the total performance difference between firmware versions.

Both stats are results of same environmental conditions and no noticeable downtime.




Results after 13 Hours on Firmware v0.9.5


Results after 13 Hours on Firmware v0.9.4

Here is the results running Jupiter model with 8 VRMs on firmware v0.9.5 for 12 Hours in an ambient room temp environment of 17°C with case cover off and house fan on top of unit.

.95 nets me around 275/257 GH/sec for a Saturn on cgminer

the crazy thing is the onboard temps don't read above 48C, and i'm not seeing ANYTHING - VRM modules, caps, etc above 55C at the most.

Makes me think they're not thermally limited for clock rate.  Maybe the buck converters can't push enough current to clock higher or there's not sufficient decoupling capacitance/other voltage issues?

Or do these have untapped potential?  I am ignorant to how these devices are clocked, if cgminer / etc software can adjust the clock frequencies or if it's a hard-soldered oscillator / clock divider or what.

301 watts at the PSU line in

this is also with the cover off, but no external fans

Update on firmware 0.9.5

Results after 8 hours.

Ambient room temp below 17 degrees celsius. Power draw at wall 580 watts. Case cover off and house fan in place.

0.93


0.95


231w vs 231w. Looks like I'm staying with 0.93 with its higher WU/m.

throw bertmod's first version of firmware on top of it to see watts

Looks like 0.93 is pushing out more shares for me.

Can't test the power differences right now as I don't have my watt-o-meter thing.

3 hour 11 min update.

I notice that the hw % is down to under 5% however I also notice that the WU/m is down from what I had from v 0.9.4

Firmware 0.9.4 averaged at just under WU8300 and HW% was at 6.5%
Firmware 0.9.5 averaged at 7800 and HW% is just under 5%

Here are the results Jupiter on firmware v0.9.5 power 580 watt @ 120v