KnCMiner Jupiter Miner First Impressions

[r3animation]

Okay I have no idea what's going on with my Saturn.

Upgraded to 0.96 yesterday (255gh/s, 3850 WU/m) and after a few hours went back to 0.94 (270gh/s, 4050 WU/m). After another few hours, went to 0.93 (270 gh/s, 4050 WU/m) and used enablecores.bin (no difference).

Today I went back to 0.96 and it has the same performance numbers (272gh/s, 4050 WU/m) as 0.93/0.94 with the usual less HW errors.

[1715555471]

1715555471

[1715555471]

1715555471

[1715555471]

1715555471

[1715555471]

1715555471

[Rampion]

Selling Jupiter hashing stable at 525GH/s, to be shipped today: https://bitcointalk.org/index.php?topic=314656

[Miz4r]

I'm getting 525.1GH/s, 76xx WU/m with 5% HW errors on 0.96 after letting it mine for 2 days straight without restarting. I'd like to think it's possible to get it up to 550GH/s and less HW errors but I'm afraid I might spend quite some time flashing firmware and hard resetting before I get the same result or better than what I have now. Anyone any ideas perhaps or should I just be happy with what I have now?

[kano]

Sitarow, from where are you getting the data in the table format (listing all your miners together)?

The blue stuff is miner.php I wrote in cgminer ...

Kano thanks again for helping with hosting.

Your welcome
And thanks for your donation as well!

[FiatKiller]

If the people from KNC read this thread I'd like to suggest adding the option to do a daily reboot and to email the status once a day, if possible.

[jelin1984]

What final is the difference  8vrms and 4 vrms board?

[Sitarow]

What final is the difference  8vrms and 4 vrms board?

Units with 8VRMs on firmware v0.9.6 take more power at the plug and generate more heat because of the fact.

Here is the results of the 4 Jupiter units after 1 day 13 hours on firmware v0.9.6

[kano]

I should add a field to do the A: to GH/s calculation ...
(miner.php allows you to calculate almost any field over a set of results or a single result line)

[Sitarow]

I should add a field to do the A: to GH/s calculation ...
(miner.php allows you to calculate almost any field over a set of results or a single result line)

That would be super!

If you can while you're at it. Make a timeline chart of performance history for GH/s HW error% and WU every hour for review!

I am also moving to a different pool to continue testing until 50btc.com pool gets back on its feet.

[kano]

I should add a field to do the A: to GH/s calculation ...
(miner.php allows you to calculate almost any field over a set of results or a single result line)

That would be super!

If you can while you're at it. Make a timeline chart of performance history for GH/s HW error% and WU every hour for review!

I am also moving to a different pool to continue testing until 50btc.com pool gets back on its feet.

Sorry - no history stored in cgminer (or the API)

I've added a button at the top 'Kano' that produces an abbreviated summary.
The "Paid GH/s" at the front is that value I was referring to.
To be exact it is:
 round(pow(2,32)*Difficulty Accepted/Elapsed/10000000)/100
Interesting that I found a bug in the 'gen' code that allows fields to be created like that
(It was replacing "Accepted" and not seeing "Difficulty Accepted" first)
I'll fix that in git shortly. done.

I've removed most of the less interesting fields so that the width is a lot less.

If you want any of the fields I removed added back just let me know.

[Sitarow]

I should add a field to do the A: to GH/s calculation ...
(miner.php allows you to calculate almost any field over a set of results or a single result line)

That would be super!

If you can while you're at it. Make a timeline chart of performance history for GH/s HW error% and WU every hour for review!

I am also moving to a different pool to continue testing until 50btc.com pool gets back on its feet.

Sorry - no history stored in cgminer (or the API)

I've added a button at the top 'Kano' that produces an abbreviated summary.
The "Paid GH/s" at the front is that value I was referring to.
To be exact it is:
 round(pow(2,32)*Difficulty Accepted/Elapsed/10000000)/100
Interesting that I found a bug in the 'gen' code that allows fields to be created like that
(It was replacing "Accepted" and not seeing "Difficulty Accepted" first)
I'll fix that in git shortly. done.

I've removed most of the less interesting fields so that the width is a lot less.

If you want any of the fields I removed added back just let me know.

Thank you Kano, this will help with the next series of performance with higher room temperatures.

here is a 10 hour result


looking good.

I plan to replace the stock intake fans on the KnC2(jupiter#2) unit with the new higher cfm 140mm fans as I have done with the other 2 miners.

So far KnC4 (Juipter #4) seems to work best from factory.

[RHA]

Luke-Jr prepared something interesting today.
The install includes development environment (a bonus for me), gets bfgminer code from github, compiles and runs.
Bfgminer is working stable here for 4 hours now.

Code:

 bfgminer version 3.3.0 - Started: [2013-10-22 10:53:46] - [  0 days 04:14:08]
 [M]anage devices [P]ool management [S]ettings [D]isplay options  [H]elp [Q]uit
 Connected to stratum.mining.eligius.st diff 64 with stratum as user 1*******
 Block: ...e4a00b9b #265334  Diff:268M ( 1.92Ph/s)  Started: [15:07:23]
 ST:2  F:0  NB:53  AS:13  BW:[125/150 B/s]  E:454.03  I:20.60mBTC/hr  BS:575k
 2/368  48.0C | 239.9/261.3/261.1Gh/s | A:15191 R:276+0(1.8%) HW:25032/2.6%
--------------------------------------------------------------------------------
 KNC 0: 47.0C | 119.1/132.4/132.8Gh/s | A: 7699 R:106+0(1.3%) HW: 8286/1.7%
 KNC 1: 41.0C | 116.4/128.9/128.3Gh/s | A: 7493 R:170+0(2.2%) HW:16753/3.5%
--------------------------------------------------------------------------------
 [2013-10-22 15:07:36] Accepted 0114f314 KNC 0fp pool 0 Diff 236/64
 [2013-10-22 15:07:36] Accepted 020d93d0 KNC 1el pool 0 Diff 124/64
 [2013-10-22 15:07:37] Accepted 00debc1f KNC 0ai pool 0 Diff 294/64
 [2013-10-22 15:07:38] Accepted 016ac81c KNC 1dq pool 0 Diff 180/64
 [2013-10-22 15:07:38] Accepted 02eec7b4 KNC 1bf pool 0 Diff 87/64
 [2013-10-22 15:07:40] Accepted 0370cbc5 KNC 1cc pool 0 Diff 74/64
 [2013-10-22 15:07:42] Accepted 039c8a68 KNC 1dm pool 0 Diff 70/64
 [2013-10-22 15:07:43] Accepted 01950766 KNC 0er pool 0 Diff 161/64
 ...

[kano]

...
Thank you Kano, this will help with the next series of performance with higher room temperatures.
...

Pity about that - they didn't implement the "API devs" or "API stats" commands for some reason.
So no device information is available via the API

miner.php has rather extensive SQL style grouping and selecting you can do on a custom page like that - so when they add that I can group the SUMMARY with the device DEVS and include device statistics (e.g. device temperature) that they provide (or do the same with STATS)

Oh and thanks again for another donation (if that was you?)
Don't worry about it though, I'm not doing extra changes to get more donations - once was more than enough for this

Edit: oh, also don't forget that "Difficulty Accepted" is random so has a sizeable variance - of course based on the share difficulty.
But that snapshot you did above (after 10hrs) pretty much says indeed KNC4 is performing better than the other 3.

[DPoS]

kano, do you mean this data?

STATUS=S,When=1382207241,Code=11,Msg=Summary,Description=cgminer 3.4.0|SUMMARY,
Elapsed=310,MHS av=450396.80,Found Blocks=0,Getworks=17,Accepted=102,Rejected=2,
Hardware Errors=7707,Utility=19.77,Discarded=26,Stale=0,Get Failures=0,Local Work=45860,
Remote Failures=0,Network Blocks=3,Total MH=139401753.5263,Work Utility=6997.99,
Difficulty Accepted=26112.00000000,Difficulty Rejected=512.00000000,Difficulty Stale=0.00000000,
Best Share=468225,Device Hardware%=17.5935,Device Rejected%=1.4183,Pool Rejected%=1.9231,Pool Stale%=0.0000|

they use this /usr/bin/api-cgminer -o

[DPoS]

Luke-Jr prepared something interesting today.
The install includes development environment (a bonus for me), gets bfgminer code from github, compiles and runs.
Bfgminer is working stable here for 4 hours now.

RHA,   I am running it too but am wondering whats the best steps on reboot when everything gets wiped out?

did you make an install file?  what about any lib files, etc?

[Sitarow]

...
Thank you Kano, this will help with the next series of performance with higher room temperatures.
...

Pity about that - they didn't implement the "API devs" or "API stats" commands for some reason.
So no device information is available via the API

miner.php has rather extensive SQL style grouping and selecting you can do on a custom page like that - so when they add that I can group the SUMMARY with the device DEVS and include device statistics (e.g. device temperature) that they provide (or do the same with STATS)

Oh and thanks again for another donation (if that was you?)
Don't worry about it though, I'm not doing extra changes to get more donations - once was more than enough for this

Edit: oh, also don't forget that "Difficulty Accepted" is random so has a sizeable variance - of course based on the share difficulty.
But that snapshot you did above (after 10hrs) pretty much says indeed KNC4 is performing better than the other 3.

Yes that was I. Know that anything you do to help knc owners to thrive is welcomed by me and many lurkers as well.

[DPoS]

i recently read someone adding 'tantalum' caps helped - is there any more info on this?

also heatsinks on the VRMS? worth trying?

heatsink if you are running pre-.95 FW and only have 4 VRM boards since those will run over spec

[RHA]

RHA,   I am running it too but am wondering whats the best steps on reboot when everything gets wiped out?

did you make an install file?  what about any lib files, etc?

No install file, I'm not that experienced with Linux and bfgminer.
I tried it just for testing. It works nice disabling no cores, but with no enabling any as well, so the net effect is worse than with cgminer.
I decided to wait until Luke makes significant improvements (e.g. fast flushing) and installing won't take an hour of opkg/compilation after every reboot.
(I would place the libraries local at /config to not let them be lost after reboot.)

[kano]

kano, do you mean this data?

STATUS=S,When=1382207241,Code=11,Msg=Summary,Description=cgminer 3.4.0|SUMMARY,
Elapsed=310,MHS av=450396.80,Found Blocks=0,Getworks=17,Accepted=102,Rejected=2,
Hardware Errors=7707,Utility=19.77,Discarded=26,Stale=0,Get Failures=0,Local Work=45860,
Remote Failures=0,Network Blocks=3,Total MH=139401753.5263,Work Utility=6997.99,
Difficulty Accepted=26112.00000000,Difficulty Rejected=512.00000000,Difficulty Stale=0.00000000,
Best Share=468225,Device Hardware%=17.5935,Device Rejected%=1.4183,Pool Rejected%=1.9231,Pool Stale%=0.0000|

they use this /usr/bin/api-cgminer -o

That's the summary that I reworked on Sitarow's output with a customsummarypage in miner.php
Yes I have a miner.php pointing at Sitarow's miners and I can see what is available.

If you look at any other version of cgminer mining on any other device, there is API 'devs' and 'stats' ... and lots of other stuff
'devs' on KFC returns an error.
On a normal device it includes extra data not in the summary for each device e.g.:

Code:

   [Enabled] => Y
   [Status] => Alive
   [Temperature] => 42.00
   [MHS 5s] => 336.12
   [Last Share Pool] => 0
   [Last Share Time] => 1382525610
   [Frequency] => 0.00
   [Diff1 Work] => 2579
   [Last Share Difficulty] => 1.00000000
   [No Device] => false
   [Last Valid Work] => 1382525610
   [Device Elapsed] => 32950

'stats' on KFC just returns what I put in there for the pools, nothing for the mining device(s).
Some mining devices have lots of interesting info in 'stats' ... e.g. one of my Klondikes:

Code:

[STATS0] =>
(
   [STATS] => 0
   [ID] => KLN0
   [Elapsed] => 19528
   [Calls] => 0
   [Wait] => 0.000000
   [Max] => 0.000000
   [Min] => 99999999.000000
   [Temp 0] => 45.87
   [Clock 0] => 300.00
   [Fan Percent 0] => 100
   [Fan RPM 0] => 0
   [Nonces / Chip 0] => 0001200 0001110 0001139 0001151 0001178 0001151 0001188 0001149 0001152 0001112 0001194 0001158 0001201 0001097 0001153 0001136 
   [Errors / Chip 0] => 0000180 0000188 0000181 0000194 0000180 0000188 0000206 0000193 0000188 0000197 0000187 0000177 0000161 0000167 0000196 0000182 
   [Hash Count] => 18109775740928
   [Error Count] => 0
   [Noise Count] => 1
   [KLine Limit] => 1024
   [KLine Used] => 1
   [KQue Delay Count] => 21434
   [KQue Delay Total] => 101968
   [KQue Delay Min] => 0
   [KQue Delay Max] => 1000
   [KQue Delay Avg] => 4.75730148
   [KQue Nonce Count] => 21434
   [KQue Nonce Total] => 19528686450
   [KQue Nonce Min] => 1999
   [KQue Nonce Max] => 10680021
   [KQue Nonce Avg] => 911107.88700196
   [WQue Size] => 9
   [WQue Cleared] => 0
   [USB Pipe] => 0
   [USB Delay] => r0 0.000000 w0 0.000000
   [USB tmo] => 161307 0
)

Neither of those return anything in the KFC about the mining device ... I guess they forgot?

[il_minatore]

Someone mounted 0.97 firmware? Impressions?