Topic: KnCMiner Jupiter Miner First Impressions - page 19. (Read 65993 times)

2 minutes before the hour I lost connection with the Jupiter. Both the pool and the cgminer status page had lost connection as well.

The web interface was responsive however ssh was not working.

After a few moments I was able to connect and cgminer logging page updated from 542Gh/s down to 511Gh/s

Their was no downtime or cgminer restart however it seemed that the beaglebone was stalled.

As you can see the system load was low. The power draw at the wall is 580 watts so I am not sure if this may have been due to the custom script to view the status of the vrm's. I may have to restart the system and not add the vrm monitor script.



I will keep an eye out and see if it does recover over the next hour to 542 GH/s

As you have said I think this had occurred. I posted the details below.

Unfortunately for this experiment I am using windows 7 pro.

still off

tomorrow morning I will go to the colocation to play some tricks

sickpig

do you power one the bad core
or still off???

are you using bertmode?

if yes pay attention because it has happened to me that the cgi that get the stats for the ASIC went in a loop consuming quite a few of resource, more to the point the script use i2c-get cli command that query the PCB about temp, consumption, enabled core etc etc 

to solve the issue i've just extracted the perl script contained in bertmode .bin (asic_status.pl) and tun it on the beaglebone.

on your desktop (I assume you're using linux) download bertmod_0_2_1_for_knc_0_94.bin  do the following:


hope it helps

Here is the same results after not moving from the update tab for over 45 mins.

KnCMiner stats after 1 hour.


I forgot to include pool results

Here is the same results after not moving from the update tab for over 14 mins.

KnCMiner stats after 45 minutes.

Thats not 100% true. The reason is that it did enable the cores if you look back.

However for some reason the system never pulled more watts as would normally be expected with cores being turned on. Perhaps those core's are not getting the adequate power needed to work?

I will do a full system power down and after 1 min power it back on. Note (Avalon's also need this as they do not turn off completely until about 30 seconds after powering down.)

Yes you are correct. However even with the firmware enabling the shut off core's the performance did not go up and the power draw on the wall was still below the 580 watts.

Also it seems the software reset did make the system pull the proper watts at the wall as it presently is at 577 watts.

However the performance started at 500 gh/s and is slowly increasing up. It presently is at 518 GH/s however I installed that custom script and it seems to not be loading the status tab with the details.

IF I remained at the status tab I noticed that the GH/s did drop down to 514 GH/s. It could have been a coincidence. However just to be sure I decided to see what load that scrypt put on the system.



Nice the load increase once I was at the status tab and it drop once i moved to the update tab?

Just wanted to share this observation.

thanks, i think because it's impossible for the firmware test a core that is disable... but if a core fails, the firmware "understand" the situation and shutdown the failed core

(sorry for my english...)

Because the firmware is not reenabling the extra cores that should be ok at the low temperatures.
The theory is that by restarting the miner, the extra cores will be used again, and performance at low temperature will be better than at high temperature.

Hypothesis:
The firmware is able to disable cores when the temperature rises, but is not able to reenable them when it falls.

wait wait wait... I don't understood (I'm not english)

-temp increase: performance low
-temp low: performance the same (but should increase)

?

thanks

If you have been following my posts. You would see that I raised the temperature in the room from sub 17 degrees Celsius (°C) to 25 degrees Celsius (°C)to test if the system would continue to operate at the optimal speed at those temps even with increased air flow using the house fan. The system started averaging around 500GH/s and the power draw at the wall was 530 watts.

After 3 hours pass at 25 degrees Celsius (°C) I then lower the room ambient temp to under 17 degrees Celsius (°C)

The core processors did become active and the VRM's did start to pull more power. However the firmware never recovered the loss performance nor did it go back to 580 watts at the wall. It just peaked at 540 watts.

The answer was that the firmware adjusted by lowering the performance to a target that would have been 500 GH/s if left alone.

However I have since issued a software restart and will see if that would be adequate to get it running optimal at the low ambient room temperature.

Software restart command sent from web interface.

Putty gave me loss connection message and this is what came up on cgminer.



At least that work was submitted.

At low temperatures you have worst performance? 

It seems that the firmware v0.9.4 is unable to restore the performance once the ambient temperature is lowered automatically.

Seems you would need to issue a restart.

I will first do a software restart from the gui to see if that works.


ASIC Board   Info presently pulling 537 watts at 120v.
0   
Temperature sensor: 39.5 C

Die ID   Cores ON   Cores OFF   %
0   48   0   100
1   48   0   100
2   47   1   97.9
3   48   0   100
DC/DC ID   ON/OFF   Status   Input Voltage   Output Voltage   Output Current
0   ON   OK   12.1 V   0.742 V   36.6 A (27.2 W)
1   OFF   OFF   12.1 V   0.701 V   0 A
2   ON   OK   12.1 V   0.754 V   38.4 A (29 W)
3   OFF   OFF   12.2 V   0.729 V   0 A
4   ON   OK   12.1 V   0.74 V   37.2 A (27.5 W)
5   OFF   OFF   12.2 V   0.716 V   0 A
6   OFF   OFF   12.2 V   0.751 V   0 A
7   ON   OK   12.1 V   0.753 V   37.3 A (28.1 W)
1   
Temperature sensor: 29.0 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   12.2 V   0.722 V   36.8 A (26.6 W)
1   OFF   OFF   12.2 V   0.714 V   0 A
2   ON   OK   12.2 V   0.733 V   36.9 A (27 W)
3   OFF   OFF   12.1 V   0.734 V   0 A
4   ON   OK   12.2 V   0.891 V   0.375 A
5   OFF   OFF   12.1 V   0.88 V   0 A
6   OFF   OFF   12.2 V   0.754 V   0 A
7   ON   OK   12.2 V   0.733 V   37.9 A (27.8 W)
2   
Temperature sensor: 28.5 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   12.1 V   0.723 V   37.1 A (26.8 W)
1   OFF   OFF   12.1 V   0.706 V   0 A
2   ON   OK   12.1 V   0.756 V   38.5 A (29.1 W)
3   OFF   OFF   12.1 V   0.735 V   0 A
4   ON   OK   12 V   0.745 V   38.1 A (28.4 W)
5   OFF   OFF   12.1 V   0.708 V   0 A
6   OFF   OFF   12.1 V   0.739 V   0 A
7   ON   OK   12.1 V   0.724 V   39.7 A (28.7 W)
3   
Temperature sensor: 30.5 C

Die ID   Cores ON   Cores OFF   %
0   48   0   100
1   48   0   100
2   41   7   85.4
3   48   0   100
DC/DC ID   ON/OFF   Status   Input Voltage   Output Voltage   Output Current
0   ON   OK   12.1 V   0.735 V   36.6 A (26.9 W)
1   OFF   OFF   12.2 V   0.708 V   0 A
2   ON   OK   12.2 V   0.747 V   37.4 A (27.9 W)
3   OFF   OFF   12.2 V   0.759 V   0 A
4   ON   OK   12 V   0.744 V   35.8 A (26.6 W)
5   OFF   OFF   12.2 V   0.723 V   0 A
6   OFF   OFF   12.2 V   0.756 V   0 A
7   ON   OK   12.2 V   0.771 V   38.6 A (29.8 W)
Total DC/DC power output: 417 W

Here is the screen shots for your review

Firmware v0.9.4 ambient room temp under 17 degrees celsius power draw at wall 540 watts @ 120v

Update.

Seems the firmware is attempting to start up the performance.


ASIC Board   Info
0   
Temperature sensor: 40.5 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   12.2 V   0.742 V   36.6 A (27.2 W)
1   OFF   OFF   12.1 V   0.701 V   0 A
2   ON   OK   12.1 V   0.738 V   38.5 A (28.4 W)
3   OFF   OFF   12.2 V   0.714 V   0 A
4   ON   OK   12.1 V   0.74 V   38.2 A (28.3 W)
5   OFF   OFF   12.2 V   0.716 V   0 A
6   OFF   OFF   12.2 V   0.751 V   0 A
7   ON   OK   12.1 V   0.737 V   37.5 A (27.6 W)
1   
Temperature sensor: 32.5 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   12.2 V   0.737 V   37.2 A (27.4 W)
1   OFF   OFF   12.2 V   0.714 V   0 A
2   ON   OK   12.2 V   0.749 V   37.3 A (27.9 W)
3   OFF   OFF   12.1 V   0.75 V   0 A
4   ON   OK   12.2 V   0.875 V   0.438 A
5   OFF   OFF   12.1 V   0.88 V   0 A
6   OFF   OFF   12.2 V   0.738 V   0 A
7   ON   OK   12.2 V   0.733 V   38.4 A (28.1 W)
2   
Temperature sensor: 32.0 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   12.1 V   0.738 V   37.4 A (27.6 W)
1   OFF   OFF   12.1 V   0.706 V   0 A
2   ON   OK   12.1 V   0.74 V   38.9 A (28.8 W)
3   OFF   OFF   12.1 V   0.72 V   0 A
4   ON   OK   12.1 V   0.729 V   38.7 A (28.2 W)
5   OFF   OFF   12.1 V   0.708 V   0 A
6   OFF   OFF   12.1 V   0.739 V   0 A
7   ON   OK   12.1 V   0.755 V   39.9 A (30.1 W)
3   
Temperature sensor: 33.5 C

Die ID   Cores ON   Cores OFF   %
0   48   0   100
1   48   0   100
2   44   4   91.7
3   48   0   100
DC/DC ID   ON/OFF   Status   Input Voltage   Output Voltage   Output Current
0   ON   OK   12.2 V   0.735 V   36.9 A (27.1 W)
1   OFF   OFF   12.1 V   0.692 V   0 A
2   ON   OK   12.2 V   0.747 V   37.7 A (28.2 W)
3   OFF   OFF   12.2 V   0.759 V   0 A
4   ON   OK   12.1 V   0.729 V   37.2 A (27.1 W)
5   OFF   OFF   12.2 V   0.707 V   0 A
6   OFF   OFF   12.2 V   0.756 V   0 A
7   ON   OK   12.2 V   0.771 V   38.7 A (29.8 W)
Total DC/DC power output: 421 W

Pulling 540 watts at the wall presently.

My room is at 81 degrees which explains your better results at 62 degrees.   Thanks for room temp info.  I might also have to go to .94 eventually for the lower wattage.  running 700 & 770 watts right now  (one board has the 8 VRMs which runs cooler which is main reason Sitarow gets lower cooling and wattage than me)

I will hope to make progress on a case mod with extra fans today

glad autumn is here, helping with lowering the temp of the room with out putting in an a/c  

Yes I think I am aware of what you are attempting.

As far as the cores are coming online. I know that just restarting cgminer will get it running at 100% and will pull 580 watts at the wall.

I just want to see if the firmware can adjust to changes in the environmental temperature with out needing any restarts.

You can tell that the core's are indeed coming back online ambient room temp is now below 18 degrees Celsius and power draw at the wall is 536 watts @ 120v.

​Note: You can see from the details that providing a low ambient temperature enables the vrm's to start using more power even if the cores were never originally disabled. You can see that occurring with ASIC Board 0 from the results with 25 degrees celsius ambient room temp to sub 18 degrees celsius ambient room temp.

Seems ASIC Module 4 VRM #4 has an issue with temperature.

I have made a note and will later on see if putting a heat sink will help mitigate that.


ASIC Board   Info
0   
Temperature sensor: 38.5 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   12.2 V   0.727 V   36.3 A (26.4 W)
1   OFF   OFF   12.1 V   0.701 V   0 A
2   ON   OK   12.1 V   0.738 V   38.4 A (28.3 W)
3   OFF   OFF   12.2 V   0.729 V   0 A
4   ON   OK   12.1 V   0.74 V   37.8 A (28 W)
5   OFF   OFF   12.2 V   0.716 V   0 A
6   OFF   OFF   12.1 V   0.751 V   0 A
7   ON   OK   12.1 V   0.737 V   37.2 A (27.4 W)
1   
Temperature sensor: 28.0 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   12.2 V   0.753 V   36.8 A (27.7 W)
1   OFF   OFF   12.2 V   0.714 V   0 A
2   ON   OK   12.2 V   0.733 V   37.1 A (27.2 W)
3   OFF   OFF   12.1 V   0.75 V   0 A
4   ON   OK   12.2 V   0.891 V   0.438 A
5   OFF   OFF   12.2 V   0.88 V   0 A
6   OFF   OFF   12.2 V   0.738 V   0 A
7   ON   OK   12.2 V   0.733 V   37.9 A (27.8 W)
2   
Temperature sensor: 27.5 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   12.1 V   0.723 V   36.8 A (26.6 W)
1   OFF   OFF   12.1 V   0.706 V   0 A
2   ON   OK   12.1 V   0.74 V   38.1 A (28.2 W)
3   OFF   OFF   12.1 V   0.735 V   0 A
4   ON   OK   12.1 V   0.729 V   37.9 A (27.6 W)
5   OFF   OFF   12.1 V   0.724 V   0 A
6   OFF   OFF   12.1 V   0.739 V   0 A
7   ON   OK   12.1 V   0.739 V   39.7 A (29.3 W)
3   
Temperature sensor: 30.0 C

Die ID   Cores ON   Cores OFF   %
0   48   0   100
1   48   0   100
2   41   7   85.4
3   48   0   100
DC/DC ID   ON/OFF   Status   Input Voltage   Output Voltage   Output Current
0   ON   OK   12.2 V   0.735 V   36.4 A (26.8 W)
1   OFF   OFF   12.2 V   0.708 V   0 A
2   ON   OK   12.2 V   0.763 V   37.2 A (28.4 W)
3   OFF   OFF   12.2 V   0.759 V   0 A
4   ON   OK   12.2 V   0.744 V   35.8 A (26.6 W)
5   OFF   OFF   12.2 V   0.723 V   0 A
6   OFF   OFF   12.2 V   0.756 V   0 A
7   ON   OK   12.2 V   0.756 V   38.6 A (29.2 W)
Total DC/DC power output: 415 W

I'll give you a chance resurrecting your 4th board try this:


press 5x the button of the controller board, wait 10 seconds,
press again 5x the button
turn your PSU off
wait 30 seconds
turn it on
goto WebUI
upgrade to 0.94 (don't ask why)
apply to reboot
goto Mining and apply your pool etc. (needed)
goto status wait till cgminer runs
goto Mining again and apply again

list your results here again