Topic: KnCMiner Jupiter Miner First Impressions - page 2. (Read 65993 times)

0.98 isn't working out for me.
Back to 0.97.

All the firmware's were run at least 24 hours without any reboot.
For that reason firmware 0.90 to 0.93 aren't listed.

--------------------	0.94	0.95	0.96	0.97	0.98
Average CGminer	270Gh/s	274Gh/s	280Gh/s	284Gh/s	278Gh/s
Average Pool	275Gh/s	263Gh/s	278Gh/s	283Gh/s	269Gh/s
Consumption	485 Watts	305 Watts	313 Watts	313 Watts	348 Watts
Temputure	54 & 62°C	42 & 47°C	42 & 47°C	40 & 44°C	44 & 50°C
Rejected	1.5%	0.22%	0.38%	0.48%	0.61%
HW	2.81%	0.79%	1.08%	1.47%	3.31%
WU	3975/m	3717/m	3933/m	4021/m	3825/m

0.94	0.95	0.96	0.97	0.98

here the link to the commit that introduce automatic en/dis core algo:

https://github.com/luke-jr/bfgminer/commit/8b690959dd4b107010217289fa69da82889e63cf

quoting comments:


 + If a core gets 10 HW errors in a row without doing any proper work
 + it is disabled for 10 seconds.
 +
 + When a core gets 10 HW errors the next time it checks when it was enabled
 + the last time and compare that to when it started to get errors.
 + If those times are close (50%) the disabled time is doubled,
 + if not it is just disabled for 10s again.



hno on kncminer irc said that there colud be some kind of tourbolence, a sort of bad interaction between good cores and bad one if you keep the latter enabled. 

anyway this morning (UTC) I've discovered that my miners were running cgminer since sunday morning due to a temporary glitch in the power line at the colo that makes the miners restart. As soon as I realized that I've checked the cores status and guess what... all cores were enable again. this is strange.

Just after that I've stopped cgminer and restart bfgminer and after 12hs I get almost the same performance.   
 

I do agree. It would be fantastic, imho


I will give it a try, but I don't know if I really have good HW. I had a dead die on a asic slots for each machine. I was able to bring it back using Enablecore.bin on 0.95. This improved my hashrate at the pools from 480 to almost 520Gh/s at the pool.

I've tried 0.97 on one of them but the higher hastrate was balanced by a way higher HW error rate (10-15%)   


wow it would be good to gain another 25Gh/s per jup.

This is the list of the thing that I've to try next:

- (SW) ckolivas cgminer version
- (SW) bfgminer automatic en/dis mechanism
- (SW) test 0.96 to really see if I could gain another 20GH/s
- (HW) mod the heatsink fans as you did (lowering them)
- (HW) add two small fans in the front to improve the intake at BBB/VRM level
- (HW) improve the VRMs airflow take into account what "The Avenger" share here:
            https://bitcointalksearch.org/topic/m.3375668
            maybe using something like trepex did here:
           



 

I haven't looked at that implementation yet

I have done similar test runs like you, and it is nice to just leave off the really bad ones and just keep on the rest, but it is also a wonder if even a 70% HW error core isn't worth just leaving on since it still provides 30% valid work

Does one terrible core effect another? Do HW errors tax the control unit?

If KNC would sell extra control units I would play around with a bad module but I am done taking downtime setting up different tests.  At some point I will RMA a few problem modules. If you have good hardware you can run .96 at very nice Gh

I put 4 of my best modules in one miner and get 545-565 on avg at btcguild

DPoS

at the moment I'm running bfgminer compiled form github repo @  0a2c45816a813bc921f69c95f6c849f643d53277 (so before luke implement the mechanisms to automatically disable and re-enable problematic cores), and this is what I do every time I need to restart  the miner sw:

- /etc/init.d/cgminer.sh start
   
  the script loop through all your cores and among other things it enables all of them

- let knc cgminer version to run for a at least 20 mins.

  that way all the faulty/too hw error prone cores will be disabled.

- stop cgminer and start bgfminer.

 
using this steps I'm able to get a lower HW % and more stable hashrate and maybe a few GH/s more i terms of performance.

As soon as I will be able to regain remote access to the miners I'll test also ckolivas cgminer version.

the honestly is nice but might scare people

can we have control on disabling/enabling cores?   bfgminer has it but seems to only makes changes on a restart (tallys and marks them, but i saw HW errors still grow on cores that were marked off during current run)

Even if we can just change the '10 HW errors in a row' to whatever we want and also if/when to try to enable cores.  I saw the code in the driver they posted

To manually get around the problem ...

From another computer that has Java and network access to the KnC, get a copy of API.class from git (if you don't already have it)
https://github.com/ckolivas/cgminer/blob/master/API.class

java API "addpool|URL,USR,PASS"

As per API-README:

Edit: assuming they haven't disabled addpool in the API

er....again my system is running version .95 there is no such button on this version to add a pool..I thought there was a way to manually set this up with .95...again at 565gh and a 5% error rate I really don't want to touch this Jupiter by taking a chance on upgrading it....that seems "ludicrous speed" for a Jupiter from what I can tell on here for a Jupiter imho...

 

Has everyone looked at the file asic_test in the /config folder? 

I have a Saturn delivered 10/25 and have not been able to run any version other than 0.95.  My file states:


ASIC board #0 has 4 DC/DC
Found a KnC miner with 192 cores
KnC ASICs: good cores: 143 from 192
KnC ASICs: good cores after re-test: 144 from 192
KnC ASICs: good cores after re-test: 144 from 192
Good cores map:
 FF FF FF FF FF FF 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00


LM75 error: Bad ID 0xFFFFFFFF
ASIC board #1 has 0 DC/DC
etc.............

Wondering if everyone's test file looks like this?

PS> Bertmod states I am only using 183 watts!

To add a pool, just click on the Mining tab next to the Status tab and click Add Pool button on the right side.

Please knc stop delivered Jupiter
Because at 2 weeks will be diff to sky

as impressive as all the above seems I still have mine on .95 ..not sure of my watt useage....it is stock plug'd it in and get the following


565gh   4.52% error rate  wu is like 7709 or some such... it slows down for knc flushwork and found block stuff ....I need to find the link to the way to add a 2nd pool to .95 does not have that option....hmm what else.... oh my temps are 40c for 2 and 27c and 31c or was last I looked...er by the by it was running at 555 the last day it has creeped up to 565gh ..also it is in a basement and the temp is 62F and I have the top off and 2 6" fans on either side along side the front blowing crossways (overlapping) moving air

anyway again impressive but I have the noob skills of a block of wood on tech....so just not comfortable with all the folks with good miners that went to .97 and went backwards...and that disabling core permanently flag on some units also was daunting would feel more comfortable with my hand held and it built into bfgminer as options or the knc version...anyway my 2c worth

anyway any enlightenment is welcome I am following this thread

and if someone has a link to  the way to add a backup pool to .95 feel free I "may or may not" be brave enough to attempt such

by the by I assume when luke jr gets his bfgminer version up and such and more polished some of this stuff will be addressed ......I only have 1 core that comes up enable unenable ... knock once in a great while...would like to see the bertmod and the enbable cores and some of the other tricks enbabled in an interface .....my miner arrived hand carried to keep it in the box in soaking wet cardboard.....((my view is prob why it works so well ....if it happened at start of trip to me they had to 'hand carry" it all the way with 2 hands .....could not play Frisbee with it and keep it in the box.)...anyway don't want any more attention from those harsh 'asic gods' as they play dice with difficulty ..drink mead and laugh down on us puny miners...

thanks in advance for any views on above

Searing

Canada is the more difficult customs with Germany
I think you might pay taxes

Did any of you have troubles or delays with customs? My Jupiter arrived in Canada on Wednesday and it is being held hostage at the airport. I contacted DHL and the only response thus far was  "Shipment was referred for examination"

Does anyone know how long this process will take? My friend told me it could take a month to clear, which would make it almost worthless to me, but I hope he is just pulling my chain.

Thanks.

I had major HW errors on 0.97 also, going back to 0.96 fixed it for me.

cgminer: 488 Gh/s
BTC Guild: 391 GH/s

Running on firmware 0.97.  Case off, stock fans, no additional cooling (on To Do list), for about 17 hours now.

Stats from cgminer:
 KnC 0:                | 485.5G/488.7Gh/s | A:3600070 R:32074 HW:1313634 WU:7444.1/m

Help!  Is the large difference between cgminer and BTC Guild because of the large number of hardware errors?  Why is this happening... did I get unlucky with my chips?

Yes case off and house fan on best results.

However replacing intake fans with high cfm fans works if you must put case cover back on.
Ambient room temp below 20 is also good.

The way I've done it is to SSH into the rig, kill the running CGMiner process that the miner autostarts, and then start up CGMiner again from the command line (inputting your info along the way). It'll show the raw hashing data so long as you maintain the SSH connection.

Keep in mind that the hashing speed shown by the pool isn't indicative of how fast your rig is actually going (thanks to variance).

how i can add manyally at cgminer at knc

restart the device when the hashing speed at pool is low!!!!!!!!!!!!!!!!!!!!!!!

any details?
 Huh Huh Huh Huh Huh Huh Huh

This jupiter is underperforming 25%
Do you see anything suspicious here?

Miner Status






Mining Status

CGMiner Status Running (pid=1760)
Last Checked Fri Oct 25 15:02:52 UTC 2013
Avg. Hash Rate 513 Gh/s
Difficulty Accepted 6,730,240
Difficulty Rejected 55,296 (0.6 %)
Hardware Errors 1,836,190 (17.6 %)
WU 7,886
Difficulty Stale 0
Network Blocks 182
Pool Rejected 0.8 %
Pool Stale 0
Best Share 83,333,458
Found Blocks 0


HW Status

ASIC slot #1 53.5 ℃
ASIC slot #2 -
ASIC slot #3 73.0 ℃
ASIC slot #4 57.5 ℃
ASIC slot #5 55.0 ℃
ASIC slot #6 -


--------------------------------------------------------------------------------

ASIC Board Status
Update ASIC Status
 
Last updated on Fri Oct 25 15:03:03 2013


ASIC Board

Info


0

Temperature sensor: 53.5 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 11.8 V 0.724 V 40.2 A (29.1 W)

1
No DC/DC detected

2
ON OK 12 V 0.741 V 40.9 A (30.3 W)

3
No DC/DC detected

4
ON OK 11.9 V 0.751 V 41.1 A (30.9 W)

5
No DC/DC detected

6
No DC/DC detected

7
ON OK 11.8 V 0.735 V 42.3 A (31.1 W)
 

2

Temperature sensor: 73.0 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 0.746 V 45 A (33.6 W)

1
No DC/DC detected

2
ON OK 11.8 V 0.746 V 44.3 A (33 W)

3
No DC/DC detected

4
ON OK 11.8 V 0.767 V 43.6 A (33.4 W)

5
No DC/DC detected

6
No DC/DC detected

7
ON OK 11.8 V 0.756 V 44.8 A (33.9 W)
 

3

Temperature sensor: 57.5 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 0.74 V 41.3 A (30.6 W)

1
No DC/DC detected

2
ON OK 12 V 0.736 V 42.7 A (31.4 W)

3
No DC/DC detected

4
ON OK 11.9 V 0.743 V 43.4 A (32.2 W)

5
No DC/DC detected

6
No DC/DC detected

7
ON OK 11.9 V 0.759 V 43.8 A (33.2 W)
 

4

Temperature sensor: 55.0 C


Die ID

Cores ON

Cores OFF

%

0 46 2 95.8
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 11.9 V 0.754 V 38.9 A (29.3 W)

1
No DC/DC detected

2
ON OK 11.9 V 0.749 V 41.5 A (31.1 W)

3
No DC/DC detected

4
ON OK 11.9 V 0.72 V 41.9 A (30.2 W)

5
No DC/DC detected

6
No DC/DC detected

7
ON OK 12 V 0.751 V 42.4 A (31.8 W)
 

Total DC/DC power output: 505 W