Pages:
Author

Topic: Hacking The KNC Firmware: Overclocking - page 57. (Read 144378 times)

newbie
Activity: 12
Merit: 0
December 24, 2013, 12:05:16 PM
DC\DC - Off after diagnostic tool for November Sad. I try enable-core, recovery and different firmware. Sad
sr. member
Activity: 386
Merit: 250
December 24, 2013, 11:21:24 AM
Need help. SD full image of October Saturn (Saturn-B52, 4 VRM)
My Bad (After wrong diagnostic tool for November). I am ready for pay if somebody do this, please:
dd if=/dev/mmcblk1 of=/dev/mmcblk0 bs=1K count=1M

Can you use the Recovery File posted on the page linked below?

https://www.kncminer.com/pages/troubleshooting

You end up with 96.1, then upgrade to FW of choice.

Not sure on instructions but basic BBB reflash is copy image to SD using PC, insert in BBB and reset.
(instructions have been posted in that other huge thread a few times)

EDIT found instructions here
http://forum.kncminer.com/forum/main-category/main-forum/11263-urgent-dl-d-the-recovery-file-sd-from-knc-support-now-how-do-i-apply-it-to-my-miner?p=11985#post11985
newbie
Activity: 47
Merit: 0
December 24, 2013, 10:47:36 AM
Sudya, have you tried installing the 0.99-tuning firmware and using the advanced tab to reset to factory defaults?

Sudya, if this didn't work, try KnC's SOS-morse-code-reset:  Press reset button 5x, wait 5sec, press button 5x

"HARD RESET
The miner is equipped with a hard reset button. This is located next to the power
input on the controller board. To perform a hard reset on the miner you must
press the button 5 times consecutively, wait 5 seconds, and then press the button a
further 5 times. Please note that the miner will not restart."

sr. member
Activity: 378
Merit: 250
December 24, 2013, 10:28:12 AM
Sudya, have you tried installing the 0.99-tuning firmware and using the advanced tab to reset to factory defaults?
newbie
Activity: 12
Merit: 0
December 24, 2013, 04:50:26 AM
Need help. SD full image of October Saturn (Saturn-B52, 4 VRM)
My Bad (After wrong diagnostic tool for November). I am ready for pay if somebody do this, please:
dd if=/dev/mmcblk1 of=/dev/mmcblk0 bs=1K count=1M

http://i.imgur.com/GhfJ0aF.png
http://i.imgur.com/X38JhI9.png
sr. member
Activity: 386
Merit: 250
December 23, 2013, 11:06:22 PM

Thanks. Btw is there any difference between installing bfgminer using your method and just installing Bertmod?
Because if there isn't then no point in doing it manually really. Is it the same version in both methods?

I plead ignorance.
I have been compiling bfgminer all along I have not kept up with bertmod.
I have heard it is possible. How? I don't know.
hero member
Activity: 812
Merit: 502
December 23, 2013, 11:00:19 PM

And at the end I changed to:

Code:
start-stop-daemon -b -S -x screen -- -S bfgminer -t bfgminer -m -d "$DAEMON" --api-listen --default-config /config/bfgminer/bfgminer.conf


My error
The "/config/cgminer.conf" at the end should remain the same.

Code:
start-stop-daemon -b -S -x screen -- -S bfgminer -t bfgminer -m -d "$DAEMON" --api-listen --default-config /config/cgminer.conf

Will edit above.
Thx for the feedback!

Smiley
Thanks. Btw is there any difference between installing bfgminer using your method and just installing Bertmod?
Because if there isn't then no point in doing it manually really. Is it the same version in both methods?
sr. member
Activity: 386
Merit: 250
December 23, 2013, 10:29:56 PM

And at the end I changed to:

Code:
start-stop-daemon -b -S -x screen -- -S bfgminer -t bfgminer -m -d "$DAEMON" --api-listen --default-config /config/bfgminer/bfgminer.conf


My error
The "/config/cgminer.conf" at the end should remain the same.

Code:
start-stop-daemon -b -S -x screen -- -S bfgminer -t bfgminer -m -d "$DAEMON" --api-listen --default-config /config/cgminer.conf

Will edit above.
Thx for the feedback!

Smiley
hero member
Activity: 812
Merit: 502
December 23, 2013, 08:21:40 PM
bitcoinorama mentioned KnC is planning to release PLL clock FW.

Great post Smiley

I think :x is easier than :wq Wink

Just tried to install and run BFGminer, but I can't run it. What am I doing wrong?

There are 5 lines in /config/bfgminer/bfgminer.sh that mention cgminer and need to be changed:

Code:
DAEMON=/config/bfgminer/bfgminer
NAME=bfgminer
DESC="Bfgminer daemon"

I assume of these 3 only the first one is important.

And at the end I changed to:

Code:
start-stop-daemon -b -S -x screen -- -S bfgminer -t bfgminer -m -d "$DAEMON" --api-listen --default-config /config/bfgminer/bfgminer.conf

And last:

Code:
do_stop() {
        killall -9 bfgminer || true

But when I try to:
Code:
/etc/init.d/cgminer.sh stop ; /config/bfgminer/bfgminer.sh restart
or
Code:
/config/bfgminer/bfgminer.sh restart
/config/bfgminer/bfgminer.sh start

Nothing happens. The last one just stays on Starting bfgminer deamon for like 15 seconds, but there are no screens later on.

What am I doing wrong?

sr. member
Activity: 322
Merit: 250
December 23, 2013, 05:28:14 PM
231 sounds pretty high.  That's 900MHz if I'm not mistaken.

How many amps does bertmod report each DCDC is putting out?  I hear that exceeding 50A each is pushing things beyond safe margins (in particular the PCB traces are at risk).

The feedback i saw from the KNC folks is that 60A is the upper end for the GE VRMs but of course they higher you go the more risk youtake
I'm pretty sure the datasheet lists 40A as the max.  But that is of course, as always, underspecified.
sr. member
Activity: 386
Merit: 250
December 23, 2013, 04:46:46 PM
bitcoinorama mentioned KnC is planning to release PLL clock FW.
You can wait for KnCMiner and save the warranty,
or not.  Smiley

orama also mentioned that KnCMiner considers 63A per GE DC/DC OK
https://bitcointalksearch.org/topic/m.4106143
Yikes!!!
That can make more heat than the ASIC can move to the heatsink IMneverHO

People wanting a different method.
This will not really lower the bar, but might help you over it.
(the sed thingy is about as easy as it can get)
More hints (a mini howto tangent)

You need vi or some other editor.
Almost every *nix system has vi on it FWIW
I only use a few commands in vi
A super abbreviated vi survival guide follows.

ESC     = escape key, I start every vi command with escape to make sure I am not in insert mode.
i       = insert
a       = append
r       = replace chr under cursor
cw      = change word under cursor
x       = delete chr under cursor
dd      = delete line cursor is on (deleted line is copied to buffer (similar to CTL-x))
Y       = yank(copy) line to buffer
P       = pop(paste) line from buffer
:       = start of command to vi (file ops etc.)
:w      = write current file to disk
:q      = quit
:wq     = write file and quit editor
:q!     = quit without saving changes(it's all fucked up start over Smiley
/       = search

Ex:
'cp /etc/init.d/cgminer.sh /config/zzz.001.sh'
'vi /config/zzz.001.sh'
hit the slash key '/' then type D1 and hit enter
Your cursor should now be on the line most reading this seek.
Yank the line to the buffer with 'Y'
Pop a copy of the buffer back into the file with 'P'
now you have 2 lines exactly the same
hit 'i' to insert text
put a '#' at the beginning of one the copies  (turn it into a comment/reminder)
hit the ESCape key to get out of insert mode
move the cursor to the other copy of the line and over the 'D' in 0xD1
hit the 'r' key and replace the 'D' with a 'C'(slower) or an 'E' or an 'F'
hit Esc then ':wq'
thats it !
It's not hard just cryptic

It is possible to edit on a Win box and xfer the edited file with 'scp'
Google 'man scp' to learn more.

bfgminer is needed to adjust all die for best results.
With temp and HW error% seperated by ASIC no need for other Status webpage.
You still need the Advanced web page though.
cgminer can run the 'result' too but not provide much feedback during discovery.
I am a fan of BOTH cgminer and bfgminer FWIW both work well on KnCMiners.

Most of you lazy fornicators will try the seemingly easier softer way without bfgminer.
You will find it is not easier and softer if you cook an ASIC or VRM.
(permanent high HW error or worse)
bfgminer will not save the ASIC but it might let you know something is amiss in time to manually intervene.
KEY WORD = might

Clock specific mini bfgminer howto:
Use your newly minted vi skills
Go to this webpage
https://github.com/luke-jr/bfgminer
Click on README.ASIC --> https://github.com/luke-jr/bfgminer/blob/bfgminer/README.ASIC
Scroll to the bottom and find and READ the KnC section
Copy the text with your mouse
Then on the miner do the following (in a SSH window)
'vi /config/BFG_build.sh'
hit the 'i' key to insert text
Right click and paste text copied from webpage

Look at the bottom of the file for 2 lines

/etc/init.d/cgminer.sh stop
./bfgminer -S knc:auto -c /config/cgminer.conf

They stop cgminer and start bfgminer as soon as it is built
You may not want to do this automatically as it does not use the mods this way.
To change this behavior,
move your cursor to the beginning of one of the lines,
hit 'i' and comment it out by inserting a '#' at the beginnig
move to the other line and hit '#' again.
Then they will look like this

#/etc/init.d/cgminer.sh stop
#./bfgminer -S knc:auto -c /config/cgminer.conf

hit Esc key
hit ':' (colon)
hit 'w'
hit 'q'
hit enter

Now run it to build bfgminer
'source /config/BFG_build.sh'

Wait (might be 10 minutes or so)

When it is done cd into the new bfgminer folder
'cd /config/bfgminer'

For our purposes we need the startup stuff cgminer uses (the file with the clock info)
'cp /etc/init.d/cgminer.sh /config/bfgminer/bfgminer.sh'
edit the file with
'vi /config/bfgminer/bfgminer.sh'
change the
DAEMON=/usr/bin/cgminer
to
DAEMON=/config/bfgminer/bfgminer

Change every instance that reads 'cgminer' so it reads 'bfgminer' instead.

EDIT*** except for the config file part.
Thx CYPER for pointing it out.
Should look like
Code:
start-stop-daemon -b -S -x screen -- -S bfgminer -t bfgminer -m -d "$DAEMON" --api-listen --default-config /config/cgminer.conf
Save the file ':wq'

Prepare your workspace
Treat every test as a pit stop with a plan A and a plan B
Test plan B first!!!!!
A good plan B might be
'/config/bfgminer/bfgminer.sh stop ; /etc/init.d/cgminer.sh restart'

open 3 SSH windows or tabs per miner
One for vi (just use Esc-colon-w and stay in the editor for fast fixes) or scp
One to start and stop modified and unmodified shell scripts (up arrow is your friend)
One to run screen -r (after you do it once just use up arrow and Enter)

One web page ALREADY open to the Advanced tuning page for temp/volt/current/watt
After starting the modified shell script this is the first place to tell how it works
A few seconds is all it takes.
3 flavors to look for
        0 or very low die (ANY die)
        10-15 watts (ANY die)
        30 watts or more (ALL die)

The first 2 require immediate action,  Plan B!
You might first QUICKLY try just restarting.

The third indicates some degree of success requiring further investigation.
Wasting power and punishing ASIC without extra hash at the pool is non productive.
Sometimes a die can show power consumption and not be contributing.

***
*** This can happen after it has been running OK for a while and be hard to find.
***

Review the pitstop plans one more time!!!

NOW stop cgminer and start bfgminer
'/etc/init.d/cgminer.sh stop'
'/config/bfgminer/bfgminer.sh restart'
(I always use restart even though it may not be running and start would be more correct, for easy reuse with the arrow keys for speed during pit stop)

In your ssh that was running the screen showing cgminer it drops back to command line

hit the up arrow and Enter(or type 'screen -r' again) and view bfgminer hashing Smiley

Suggest first run with bfgminer use unmodified clocks
If you got this far you are ready to add a section for each ASIC and die to COPY_OF.bfgminer.sh

!!!Also note!!!
The miner has a DC/DC monitor that is occasionally unhelpful.
(possibly because you are stopping cgminer and starting bfgminer when it looks)
If it does this it will start cgminer.
This can occasionally happen when restarting bfgminer.
The solution is '/etc/init.d/cgminer.sh stop ; /config/bfgminer/bfgminer.sh restart'
(Advanced web page shows it running but bfgminer says nada or 'screen -r' says already running or somesuch)

MAKE NOTES!!!
16 different die in one box can be confusing.

Other fun toys are in the bfgminer(or a cgminer) source directory too.
read the file README.RPC
'less /config/bfgminer/README.RPC'
then look at
miner.php
^^^^^^^^^
(I run this on my PC,
it collects all miners into one place,
and shows more status of each core)

also look at
api-example
api-example.php
api-example.py

A lot can be done with just api-example (and grep)

Don't reboot, rarely needed, else rebuild bfgminer again after.
Just 'q'uit and run unmodified
'/etc/init.d/cgminer.sh restart'
instead.
(ways around this limitation exist but are beyond the already large scope of this post)

Hope this helps someone get over the hump and wake a sleeper and possibly into new GH territory.

Smiley


newbie
Activity: 32
Merit: 0
December 23, 2013, 03:51:59 PM
Thxs to this thread and alittle help on IRC...I turned up my jup. to around 630-650gh/s on poolside Smiley thxs tolip great job and my HW errors down to .5%
Not enough information..That doesn't help anyone. What do your config look like? 4/8VRM? Etc.
Why not share information? Sad
It's not that hard really, consider myself a noob and sid not have a lot of trouble.
Changed the clock to 231 (plenty of how-to above).
Performance is steady 650-670 on pool, have 3 x8-vrm boards and 1x 4-vrm (RMA-d). Temps 43 (intake side boards) to 55 degrees, I lowered the fans for more flow over the boards.
Power usage (bertmod) from 451 W to 644 W (speed from 562 to 678 In cgminer), hw 3.5% (tightened the voltages a bit which increased hw-errot but reduced power/avoid it gets to hot.

Asic chips can handle more if I base it on the heat coming of them but don't know any higher clock frequencies as 231 ( besides that the vrm's are around their 40W limit already). Guess more juice can be pumped out if 8 vrms can be activated again!
Yea, but that should be a Oct 8VRM miner with the availability if 99-tuning (=voltage regulations).
I wish there was a firmware with that to the nov as well.

So, has anyone tried this on nov. unit except from me? Those negative results (340+ -> 333Gh/s at best) is making me sad Sad
hero member
Activity: 812
Merit: 502
December 23, 2013, 03:18:23 PM
I have a 3 Module Saturn that is running at 500Gh using 211 with no issues.

231 gets me about 530-540 but until the temp in my building drops into the 40's I will leave it as is.

500GH/s I can believe, but 540 is a science fiction, sorry.
If we are talking about the average speed reported by the pool.
hero member
Activity: 854
Merit: 500
December 23, 2013, 03:16:09 PM
I have a 3 Module Saturn that is running at 500Gh using 211 with no issues.

231 gets me about 530-540 but until the temp in my building drops into the 40's I will leave it as is.
hero member
Activity: 812
Merit: 502
December 23, 2013, 03:07:13 PM
231 sounds pretty high.  That's 900MHz if I'm not mistaken.

How many amps does bertmod report each DCDC is putting out?  I hear that exceeding 50A each is pushing things beyond safe margins (in particular the PCB traces are at risk).

The feedback i saw from the KNC folks is that 60A is the upper end for the GE VRMs but of course they higher you go the more risk youtake

Can you really overclock the November ones above 670GH/s?
hero member
Activity: 854
Merit: 500
December 23, 2013, 03:05:34 PM
231 sounds pretty high.  That's 900MHz if I'm not mistaken.

How many amps does bertmod report each DCDC is putting out?  I hear that exceeding 50A each is pushing things beyond safe margins (in particular the PCB traces are at risk).

The feedback i saw from the KNC folks is that 60A is the upper end for the GE VRMs but of course they higher you go the more risk youtake
hero member
Activity: 812
Merit: 502
December 23, 2013, 02:57:03 PM
231 sounds pretty high.  That's 900MHz if I'm not mistaken.

How many amps does bertmod report each DCDC is putting out?  I hear that exceeding 50A each is pushing things beyond safe margins (in particular the PCB traces are at risk).

Check this post: https://bitcointalksearch.org/topic/m.4106607
legendary
Activity: 966
Merit: 1000
December 23, 2013, 02:33:40 PM
231 sounds pretty high.  That's 900MHz if I'm not mistaken.

How many amps does bertmod report each DCDC is putting out?  I hear that exceeding 50A each is pushing things beyond safe margins (in particular the PCB traces are at risk).
newbie
Activity: 31
Merit: 0
December 23, 2013, 02:11:25 PM
Thxs to this thread and alittle help on IRC...I turned up my jup. to around 630-650gh/s on poolside Smiley thxs tolip great job and my HW errors down to .5%
Not enough information..That doesn't help anyone. What do your config look like? 4/8VRM? Etc.
Why not share information? Sad
It's not that hard really, consider myself a noob and sid not have a lot of trouble.
Changed the clock to 231 (plenty of how-to above).
Performance is steady 650-670 on pool, have 3 x8-vrm boards and 1x 4-vrm (RMA-d). Temps 43 (intake side boards) to 55 degrees, I lowered the fans for more flow over the boards.
Power usage (bertmod) from 451 W to 644 W (speed from 562 to 678 In cgminer), hw 3.5% (tightened the voltages a bit which increased hw-errot but reduced power/avoid it gets to hot.

Asic chips can handle more if I base it on the heat coming of them but don't know any higher clock frequencies as 231 ( besides that the vrm's are around their 40W limit already). Guess more juice can be pumped out if 8 vrms can be activated again!
newbie
Activity: 32
Merit: 0
December 23, 2013, 02:59:57 AM
Thxs to this thread and alittle help on IRC...I turned up my jup. to around 630-650gh/s on poolside Smiley thxs tolip great job and my HW errors down to .5%
Not enough information..That doesn't help anyone. What do your config look like? 4/8VRM? Etc.
Why not share information? Sad
Pages:
Jump to: