Pages:
Author

Topic: Braiins OS & Braiins OS+ custom ASIC firmware: optimize performance & efficiency - page 26. (Read 45390 times)

legendary
Activity: 2030
Merit: 1569
CLEAN non GPL infringing code made in Rust lang
Would it be possible (maybe in future releases) to control the startup fan speed?

And instead of using no-sensor-scan to ignore all temperature sensors, would it be possible to just ignore the sensors on boards/chains with problems?

I asked about this, and there is good news for you:

Quote from: Artemis3
"would it be possible to just ignore the sensors on boards/chains with problems?"

I think this feature request would be to disable sensors per board (chain) instead of the whole miner.

Quote from: Jan Čapek
that's implemented for the upcoming release

Not about the fan of course. There is nothing you can do when you are powering up a fan you get full power to it, until the controller boots and orders it to slow down. The only way to solve this would be with a hardware solution, such as feeding 7v instead of 12v for 1min or such...
newbie
Activity: 16
Merit: 10
Would it be possible (maybe in future releases) to control the startup fan speed?

And instead of using no-sensor-scan to ignore all temperature sensors, would it be possible to just ignore the sensors on boards/chains with problems?
legendary
Activity: 2030
Merit: 1569
CLEAN non GPL infringing code made in Rust lang
To upgrade the Braiins OS firmware to the latest version, upgrade the package called "firmware" (yes, thats its name). You can do that from the web UI (first refresh packages), or from SSH with:

Code:
opkg update && opkg upgrade firmware

You can follow this procedure every time there is a new Braiins OS version available.
full member
Activity: 686
Merit: 140
Linux FOREVER! Resistance is futile!!!
Can I upload this through the braiins firmware upgrade gui to refresh braiins?

braiins-os_am1-s9_ssh_2019-02-21-0-572dd48c.tar.bz2

To use web upgrade you should use web image

braiins-os_am1-s9_web_2019-02-21-0-572dd48c.tar.gz
hero member
Activity: 770
Merit: 523
Can I upload this through the braiins firmware upgrade gui to refresh braiins?

braiins-os_am1-s9_ssh_2019-02-21-0-572dd48c.tar.bz2
newbie
Activity: 72
Merit: 0
Hmm i would guess /etc/init.d/cgminer stop

But why?

Im playing with cgminer source code and need to replace binary. So first I have to stop it.
member
Activity: 71
Merit: 17
The error you posted it's trying to upload the s9 firmware, use the t1 firmware instead.

LOL, im so dumb, thanks for the advice, now seems that it goes, is "Restarting" for a couple minutes.
sr. member
Activity: 463
Merit: 309
The error you posted it's trying to upload the s9 firmware, use the t1 firmware instead.
member
Activity: 71
Merit: 17
Im having this trouble when putting de fw in a Dragonmint T1 with G19 board on windows.

Code:
Preparing system for backup...
Traceback (most recent call last):
  File "upgrade2bos.py", line 185, in
    main(args)
  File "upgrade2bos.py", line 90, in main
    if not platform.backup_firmware(args, ssh, backup_dir, mac):
  File "G:\braiins\braiins-os_am1-s9_ssh_2019-02-21-0-572dd48c\upgrade\platform.py", line 77, in backup_firmware
    ssh.run('/etc/init.d/bmminer.sh', 'stop')
  File "G:\braiins\braiins-os_am1-s9_ssh_2019-02-21-0-572dd48c\upgrade\ssh.py", line 228, in run
    self._check_exit_status(cmd, stdout, stderr)
  File "G:\braiins\braiins-os_am1-s9_ssh_2019-02-21-0-572dd48c\upgrade\ssh.py", line 159, in _check_exit_status
    raise CalledProcessError(returncode, cmd, stdout, stderr)
subprocess.CalledProcessError: Command '/etc/init.d/bmminer.sh stop' returned non-zero exit status 127.

Then nothing happened and the fw keeps the same, someone can help me?
legendary
Activity: 2030
Merit: 1569
CLEAN non GPL infringing code made in Rust lang
Hi Guys

Silly question. How can I turn off cgminer in S9 BraiinOS while leave all system running.

Thanks!

Hmm i would guess /etc/init.d/cgminer stop

But why?
newbie
Activity: 72
Merit: 0
Hi Guys

Silly question. How can I turn off cgminer in S9 BraiinOS while leave all system running.

Thanks!
hero member
Activity: 770
Merit: 523
Had an xilinx board with bitmain may 3 s9 firmware. Tried to install braiins with no luck. Then I flashed bitmain dec 5 s9 firmware on it and THEN installed braiins and I am good to go.

I have a few  t9+ miners I was going to see if I could put a braiins flashed s9 xilinx board in it instead, have not gotten around to it yet. I know I probably need to throw a few more commas in the cgminer.conf for the mhz and the voltage. Next on my list.



Can confirm this is working, have done the same process to a T9 (NOT T9+) that was shutting down because of overheating.

Flashed the T9 to the S9 Dec 5 firmware, then flashed brains OS over the top. Boots and runs, and you can control the chip speed and voltage, but as wolfen said you get no temp readout and the fans run at full speed.

Would be nice to get temps, but I suspect the braiins guys would need a T9 to test/tweak.

This particular machine has been problematic, and prone to overheating, so I wanted to underclock and undervolt it, but bitmain won't release AB firmware for the T9, and the old trick of going into minerAdvanced.cgi never did anything, it appears to be working for this one. I have it at 600Mhz and 8.5V

If I stick one 63 chip board in with two 57 chip boards I get one chip temp reading on the 63 chip board. Then it uses the one 63 chip board temp
to control the fan. I have had stock 550mhz miner chip temps run at 115C all day. I have a few miners at 105C dangerous temp now so they won't shut off too early.

Some 63 chip hash boards have no temp either, I just make sure I have ONE hash board with temp. I just pulled a board out of my dead pile that had no temp. Works fine with no-sensor-scan in a single board miner. On bitmain firmware it did not run.

Nice to be able to mix and match with braiins.
full member
Activity: 261
Merit: 100
Hi, would you consider preparing the  Braiins OS firmware for the T9+ ?
legendary
Activity: 4256
Merit: 8551
'The right to privacy matters'
[...]

ASIC boards vary like 20% to 0 %

You have good settings for that set of boards.

But your other 19 may be as much as 10% different.

Use those settings one one s9 at a time and try to tweak it.

If you are at 8.7 and 700 try 8.6 and see if any asic boards can hold it at 8.6
Also try at 8.8 so this means 3 volt settings on three cards or 9 settings just for freq 700

I only have two braiins s9s as I simply don’t have time to flash them.
The two I have were eBay purchases. I got lucky to find them.

I have 69 units
s9
s15
t15
m10
a721
A921
T2turbo24
S7ln
L3
L3+
And 1 simple mining rig with 4 GPUs

I am a 75-90 minute trip from location. I have not fully set it up. We will expand more and eventually dump the 22 s9s.  My power deal is good so I did not put a proirity on flashing the other s9s.
legendary
Activity: 3234
Merit: 1220
Hmm this reminds me of the R4, but just like the T9s, some R4s have different controllers where everything just breaks. Still good finding.
Did you try forcing the fan speeds slower?

Well for me, fan speed isn't an issue with this miner. And as I can't see what its temps are I'm just running them at 100% rather than risk it.

Also I think one fan needs replacing as its indicating over 9000rpm, but I doubt it's going to happen given its location and difficulty accessing it.
legendary
Activity: 2030
Merit: 1569
CLEAN non GPL infringing code made in Rust lang
Can confirm this is working, have done the same process to a T9 (NOT T9+) that was shutting down because of overheating.

Flashed the T9 to the S9 Dec 5 firmware, then flashed brains OS over the top. Boots and runs, and you can control the chip speed and voltage, but as wolfen said you get no temp readout and the fans run at full speed.

Would be nice to get temps, but I suspect the braiins guys would need a T9 to test/tweak.

This particular machine has been problematic, and prone to overheating, so I wanted to underclock and undervolt it, but bitmain won't release AB firmware for the T9, and the old trick of going into minerAdvanced.cgi never did anything, it appears to be working for this one. I have it at 600Mhz and 8.5V

Hmm this reminds me of the R4, but just like the T9s, some R4s have different controllers where everything just breaks. Still good finding.
Did you try forcing the fan speeds slower?
legendary
Activity: 3234
Merit: 1220
Took a t9 with 3 57 chips boards.
Flashed controller to s9 dec5. Then flashed to braiins. Boots up and runs.
No temps show but is appears to be running well. Fans run at full speed.

Can confirm this is working, have done the same process to a T9 (NOT T9+) that was shutting down because of overheating.

Flashed the T9 to the S9 Dec 5 firmware, then flashed brains OS over the top. Boots and runs, and you can control the chip speed and voltage, but as wolfen said you get no temp readout and the fans run at full speed.

Would be nice to get temps, but I suspect the braiins guys would need a T9 to test/tweak.

This particular machine has been problematic, and prone to overheating, so I wanted to underclock and undervolt it, but bitmain won't release AB firmware for the T9, and the old trick of going into minerAdvanced.cgi never did anything, it appears to be working for this one. I have it at 600Mhz and 8.5V
legendary
Activity: 2030
Merit: 1569
CLEAN non GPL infringing code made in Rust lang
It is looking ok. You have to repeat this calibration with each of your 20 units, all 60 Asic boards separately. Don't just blind copy the settings because they are not identical, every Asic board responds differently. Make sure to inspect your PSU and connections for excess heat so they don't die prematurely.

Ensure good airflow around the miner.

Overclocking tends to shorten the lifetime or the miners. Make sure you are fully aware that some might get damaged by doing this. It is your own responsibility to be careful with each voltage increase.
member
Activity: 126
Merit: 10
I have 1600w PSU models for my antminers S9.

The electricity is not a problem for me as I host them in a factory where they pay monthly for 200KW and they don't use it all.

I give 40% of the income to the factory for the electricity and security of the miners. So THS is very important for me as more THS i get from miners as better income.

The factory contracts ends in 2020 november then I will have to stop the miners and maybe sell or not use them anymore i would like to get more from them asap.

If i pay for electricity and security is a must in my country for 20 miners it's not worth it.

I was able to get same rate as ideal and chip rate.

I did add a print screen to see how did i configure it.

The temp is way better now then it was before.

Please check the picture and let me know is it ok if I leave this configuration.

I'm planing to install  this to all my 20 Antminer S9 and have the same configuration.





Update:

The HW errors did drop please check the screenshot.

legendary
Activity: 2030
Merit: 1569
CLEAN non GPL infringing code made in Rust lang
It doesn't work that way, you just can't blindly input values others use and expect the same results.

Assuming you have a strong PSU (ie not the 1600w model, but the 1800w model or bigger), you:

Slowly rise the speed AND WAIT to see if your "Chip rate" and "Ideal rate" match, and don't see many errors. Your capture is already showing two things: 1st you are using 725mhz not 700mhz. 2nd your second and third chain (asic boards) are not making it, that is the moment you rise the voltage to them and try again. Ie. from 8.7 to 8.8. You might need to keep going up (individually), but if it does not work after going 9v or so then slow it back down to 700mhz until the rates match again.

ONCE you get matching rates you should try reducing the voltages one step at a time until you see the rates mismatching again, then you know whats the lowest "safe" voltage at a given speed for that board.

Every Asic board (chain) behaves differently.

Also this is completely unrelated to fans, the more you overclock the more hot it becomes and they will tend to go 100% and stay there (as your capture is already showing). You might also want to know that PSUs going beyond 60%~80% their capacity are less efficient, so you are wasting more power for th when using smaller capacity PSUs. Oh and you don't want to overload your PSU...

You should always aim for the best efficiency not the max TH/s.
Pages:
Jump to: