KnCMiner Jupiter Miner First Impressions - page 16.

Sitarow

legendary

Activity: 1792

Merit: 1047

40 Hour Update KnCminer running with open case and house fan in ambient room temp of 21°C

Firmware V0.9.5 Results

The first 24 hours had an ambient room temp of 17°C.

The last 16 hours had an ambient room temp of 21°C.

You can make reference to the earlier post for 24 hour and 12 hour results and compare the difference in performance.

I will now place the case cover over the unit and keep the room temp at 21°C to see if the air intake stock fans are adequate to ventilate the heat out of the case.

If not I will replace the intake stock fans with those found in the photo's/links posted early in this thread.

Edit: forgot to lower the power draw on the photos from 580 watts to the new 573 watts.

AFox

hero member

Activity: 539

Merit: 517

Quote from: r3animation on October 12, 2013, 07:57:43 PM

You might want to give 0.93 a try.

I think that gives the best performance so far for Saturns.

I'm trying it out now, 0.93 seems good. I'll add it to my stats after running it 24h.

sickpig

legendary

Activity: 1260

Merit: 1008

jelin1984

legendary

Activity: 2408

Merit: 1004

exactly the same and mine

the faulty core
have stickers number 3

but i can run it only with 0.91

mininganon

newbie

Activity: 28

Merit: 0

An Update on my Jupiter: After taking some advice on the KNC forum I updated to .95, rebooted then did a "hard reset" as outlined in the KNC manual. I saw .95 wasn't working out and switched to .94. I'm now stable at .94 with a cool room where before the hard reset it would drop ghs to unacceptable levels. I'm getting this on Eligius.st right now
3 hours   473.45 Gh/s   1190528
22.5 minutes   475.64 Gh/s   149504
256 seconds   551.80 Gh/s   32890
128 seconds   523.45 Gh/s   15600

My only issue now is this module marked with a "3" sticker on it. My other three modules have "1" and are fine.

Temperature sensor: 31.5 C
Die ID    Cores ON    Cores OFF    %
0   48   0    100
1   0   48    0
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    0.744 V    34.9 A (26 W)
1    OFF    OFF    11.9 V    0.712 V    0 A
2    ON    OK    11.9 V    0.815 V    15.1 A (12.3 W)
3    OFF    OFF    11.9 V    0.825 V    0 A
4    ON    OK    12 V    0.745 V    38.6 A (28.8 W)
5    OFF    OFF    11.9 V    0.695 V    0 A
6    OFF    OFF    11.8 V    0.739 V    0 A
7    ON    OK    11.9 V    0.732 V    40.2 A (29.4 W)

DPoS

sr. member

Activity: 462

Merit: 250

.93 is best for making money unless you have a beat up miner that cant escape FPGA error loops

.94 will let you play in the same ballpark as .93 but it lets you slide downhill if you don't watch it.
I created a script to restart cgminer on a timed loop. (they didn't install crontab) I had it set for 30 mins but extended it to 1 1/2 hours after I cooled the chips more.

Sadly, changing my one miner to .95 degraded it but I have forced it to hash around ~420 with .94 and auto restarts.
The other miner is still ~520 running .93 solid and hasn't even restarted cgminer in days

I am starting to think I might not have a dead VRM but perhaps a whole die of 48 cores is dead so the VRM shows < 1 amp.

I vented enough to KNC via email and wonder if they will say anything about reducing the VRMs to 4 before actually testing the result of such a blatant bait and switch reckless move.

I am working on redesigning the case... I have a few ideas to explore. Just about anything is better than what they slapped together for jupiters

and here's my pain summed up in a picture:

r3animation

hero member

Activity: 756

Merit: 500

You might want to give 0.93 a try.

I think that gives the best performance so far for Saturns.

Sitarow

legendary

Activity: 1792

Merit: 1047

AFox

hero member

Activity: 539

Merit: 517

Sitarow

legendary

Activity: 1792

Merit: 1047

Quote from: OC19850520 on October 12, 2013, 05:36:32 PM

Quote from: Sitarow on October 12, 2013, 03:55:06 PM

You can tell that over the course of the 24 hours firmware 0.9.5 gave more stable speeds and less hardware error %.
However firmware v0.9.4 gave more submitted shares and a WU rate of over 8200 steady.

Both firmware power usage fluctuated between 578 and 580 watts.

So more shares is better. v0.9.4 wins, or I missing something? How much luck plays role here?

Thus for 0.9.4 does win only if you can lower your ambient room temp under 21°C and know that if the temp goes up the system will throttle and even if the temp does go back down. You will have to issue a software restart in order to gain back the performance lost.

Also if you have the ability to see what the power draw is at the wall. If it is over 600 watts then you may have to contend with the reality that one or all of your ASIC boards having VRM issues.

Lowering temps seems to solve that problem.

However I will be conducting the next test by raising the ambient temp in the room to 25°C.

This is to test what the system does over the next 12 hours.

OC19850520

full member

Activity: 189

Merit: 100

Quote from: Sitarow on October 12, 2013, 03:55:06 PM

You can tell that over the course of the 24 hours firmware 0.9.5 gave more stable speeds and less hardware error %.
However firmware v0.9.4 gave more submitted shares and a WU rate of over 8200 steady.

Both firmware power usage fluctuated between 578 and 580 watts.

So more shares is better. v0.9.4 wins, or I missing something? How much luck plays role here?

Fishbones

full member

Activity: 122

Merit: 100

Quote from: Sitarow on October 12, 2013, 03:55:06 PM

You can tell that over the course of the 24 hours firmware 0.9.5 gave more stable speeds and less hardware error %.
However firmware v0.9.4 gave more submitted shares and a WU rate of over 8200 steady.

Both firmware power usage fluctuated between 578 and 580 watts.

Thanks for the detailed analyses Sitarow.
Could you also please forward KnC Support this info and link to this thread, especially regarding the lower WU with 0.95. Seems most 4VRM and 8VRM units are experiencing this. It's costing me 0.05BTC/Block Cry

Keep up the good work.

Cheers.

Sitarow

legendary

Activity: 1792

Merit: 1047

Quote from: viriat0 on October 12, 2013, 03:56:09 PM

WOW nice setup!

My friend accept donations

good luck

Thank you.

However my motivation is to help those who have been having troubles and those that are just now getting their units.
If you wish to donate then perhaps we as a community can raise the funds so the cgminer team can get ahold of a saturn and make our lives easier

viriat0

hero member

Activity: 1008

Merit: 501

WOW nice setup!

My friend accept donations

good luck

Sitarow

legendary

Activity: 1792

Merit: 1047

Update on 8 VRM Jupiter running firmware 0.9.5 Setup with open case with added house fan providing airflow over all components.

Stable environment conditions with an ambient room temp of 17°C / 62.6°F

24 hour results

Detailed Results Firmware v0.9.5 over 24 hours.

Compare with Results for Firmware v0.9.4 over 24 hours.

You can tell that over the course of the 24 hours firmware 0.9.5 gave more stable speeds and less hardware error %.
However firmware v0.9.4 gave more submitted shares and a WU rate of over 8200 steady.

Both firmware power usage fluctuated between 578 and 580 watts.

jelin1984

legendary

Activity: 2408

Merit: 1004

do you beleive that is vrm that cause the problem?

what knc tell you???

why knc does not make any custom firmware for that problem?

also knc had tell me that if i will run the firmware for 1 day will fix the problem
do you have it?
run 24 hours the same firmware?

DPoS

sr. member

Activity: 462

Merit: 250

Quote from: jelin1984 on October 12, 2013, 01:44:55 PM

because with firmware 0.95

1 chip of my jupiter is total off

i can not undresatnd why but total off

it is a shame that you have 8 VRMs and not one of the extra 4 are used to failover for the failed one. I thought that was the point

Since you have some amps on the bad VRM the modded firmware is able to detect you have all 48 cores off.

You do have good readings on the rest. (not many cores off)

a bad VRM can create the FPGA error loops which will cripple a board. Sometimes the whole miner (like mine if I try .93 now)

Do you see any FPGA errors when running?

If you get best results with .91 then stick with that

I miss the performance of .93 since it would restart cgminer if too many HW errors, but .94 seems to just let the miner slide down to bad performance. I can start around ~440Gh with my bad miner but after a few hours it is down to ~400Gh

I will probably put a restart command in the cron for every hour or 30 minutes if that proves to be optimal.

jelin1984

legendary

Activity: 2408

Merit: 1004

because with firmware 0.95

1 chip of my jupiter is total off

i can not undresatnd why but total off

RoadStress

legendary

Activity: 1904

Merit: 1007

Why are you people still using anything other than 0.95? It's the best firmware today!

jelin1984

legendary

Activity: 2408

Merit: 1004

with firmware 0.90-0.91

the wrong board
have only 48 off

with firmware 0.92-0.95
the wrong board have ALL THE CORE OFF

why that happen??? Huh

Topic: KnCMiner Jupiter Miner First Impressions - page 16. (Read 65993 times)