L3+/L3++ voltage tuning script

[psycodad]

Since jstefanop was so kind to release his demo code for voltage setting on the L3+ we could finally undervolt our L3+ and save some of those precious watts.

Though the process of tuning every hash board is time consuming and boring (at least after the third L3+) and I soon wished there was a way to automate this trial and error process similar to what General Tarkin did on the Titans.

So I hotglued some python script that does the whole process automatically and does not require you to install jstefanops binary on every single miner. But you need a Linux (MacOS might work) machine that can reach the miners you want to tune over the network.

Note: Just to be very clear: The script needs to be run from a different computer than your miner!

If you are willing to try this alpha software, head over to github and checkout my bitmain-tools repo:

https://github.com/psycodad-bct/bitmain-tools

Usage in the simplest form is just:

Code:

git clone https://github.com/psycodad-bct/bitmain-tools
cd bitmain-tools/scripts
./l3plus_autotune.py -i <your-miner-ip>

All options are printed if you supply no arguments to the script:

Code:

Usage:
./l3plus_autotune.py -i <ip>|--minerip=<ip> [OPTIONS]

Options:
 -p <adminpass>			admin password if not set to 'admin'
 --password=<adminpass>
 -s <chain1[,chain2]>		skip one or more chains
 --skip <chain1[,chain2]>

Examples:
Tune miner on 10.10.10.33, use '1234' as admin password and skip tuning chain 2 and 3:
./l3plus_autotune.py --minerip 10.10.10.33 -p 1234 --skip 2,3
Default usage :
./l3plus_autotune.py -i 10.10.10.33

The script may take anywhere from 30min to several hours or can even refuse to complete if your miner is too unstable. If that happens you can try to use the --skip option to skip tuning one or more chains/hashboards and set them to the best possible value manually.

I did this to save time on tuning and re-tuning my L3+ miners and thought I'd share it here.

Disclaimer:
This script has at best pre-alpha-quality (or even less) and comes as-is with no warranties at all. I have tested it heavily and to the best of my knowledge it should do no harm, but it has obviously when playing with voltages the potential to damage your miner and even if not it will probably void your Bitmain warranty.

Use completely at your own risk!

If in doubt, use the source Luke!

[arnold_mad]

Hi !

Tried to use it on one of my miners and got the following log:

/config/sv binary not found on target miner, installing it first:
MD5sum [113ad2c06daac293386e28807ea35671] matches, good.
Binary /config/sv successfully installed.
Traceback (most recent call last):
  File "./l3plus_autotune.py", line 545, in <module>
    current_stats = get_minerstats(miner_ip, port=4028)
  File "./l3plus_autotune.py", line 95, in get_minerstats
    resp = json.loads(json_resp)
  File "/usr/lib/python2.7/json/__init__.py", line 339, in loads
    return _default_decoder.decode(s)
  File "/usr/lib/python2.7/json/decoder.py", line 364, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
  File "/usr/lib/python2.7/json/decoder.py", line 380, in raw_decode
    obj, end = self.scan_once(s, idx)
ValueError: Expecting , delimiter: line 1 column 236 (char 235)


can you help ?
the script is running on a RPI3

[rs1x]

I am a little green in the ASIC world. What kind of tuning would this do. Apply the best bang for your buck voltage  to each of the cards ?  Does it go for higher hash power, etc ?

Sorry for the newb question. 

[psycodad]

Hi !

Tried to use it on one of my miners and got the following log:

/config/sv binary not found on target miner, installing it first:
MD5sum [113ad2c06daac293386e28807ea35671] matches, good.
Binary /config/sv successfully installed.
Traceback (most recent call last):
  File "./l3plus_autotune.py", line 545, in <module>
    current_stats = get_minerstats(miner_ip, port=4028)
  File "./l3plus_autotune.py", line 95, in get_minerstats
    resp = json.loads(json_resp)
  File "/usr/lib/python2.7/json/__init__.py", line 339, in loads
    return _default_decoder.decode(s)
  File "/usr/lib/python2.7/json/decoder.py", line 364, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
  File "/usr/lib/python2.7/json/decoder.py", line 380, in raw_decode
    obj, end = self.scan_once(s, idx)
ValueError: Expecting , delimiter: line 1 column 236 (char 235)


can you help ?
the script is running on a RPI3

Thanks for testing the script.

It seems to fall over while parsing the cgminer json output.
I am a bit at loss, would need to see the raw json output from cgminer.
EDIT:
You could try to add a print statement just before the line it falls over, i.e. before line 95 add
  print json_resp
and post or pm the output.
I have pushed an update to github which should print the json data so we can see whats wrong or where it falls over.
Please update with git pull and post or pm me the output.

Also, are you running the latest Bitmain firmware 1.0.1.3 ?

[psycodad]

I am a little green in the ASIC world. What kind of tuning would this do. Apply the best bang for your buck voltage  to each of the cards ?  Does it go for higher hash power, etc ?

Sorry for the newb question. 

Basically you set a frequency for your L3+ and then this script tunes the miner to the best voltage settings in terms of optimal error_ratio/power-consumption per board.

[arnold_mad]

Hi !

Tried to use it on one of my miners and got the following log:

/config/sv binary not found on target miner, installing it first:
MD5sum [113ad2c06daac293386e28807ea35671] matches, good.
Binary /config/sv successfully installed.
Traceback (most recent call last):
  File "./l3plus_autotune.py", line 545, in <module>
    current_stats = get_minerstats(miner_ip, port=4028)
  File "./l3plus_autotune.py", line 95, in get_minerstats
    resp = json.loads(json_resp)
  File "/usr/lib/python2.7/json/__init__.py", line 339, in loads
    return _default_decoder.decode(s)
  File "/usr/lib/python2.7/json/decoder.py", line 364, in decode
    obj, end = self.raw_decode(s, idx=_w(s, 0).end())
  File "/usr/lib/python2.7/json/decoder.py", line 380, in raw_decode
    obj, end = self.scan_once(s, idx)
ValueError: Expecting , delimiter: line 1 column 236 (char 235)


can you help ?
the script is running on a RPI3

Thanks for testing the script.

It seems to fall over while parsing the cgminer json output.
I am a bit at loss, would need to see the raw json output from cgminer.
EDIT:
You could try to add a print statement just before the line it falls over, i.e. before line 95 add
  print json_resp
and post or pm the output.
I have pushed an update to github which should print the json data so we can see whats wrong or where it falls over.
Please update with git pull and post or pm me the output.

Also, are you running the latest Bitmain firmware 1.0.1.3 ?

I'm running the custom blissz firmware. here the json output:

Failed to decode json reply:
Expecting , delimiter: line 1 column 236 (char 235)

{"STATUS":[{"STATUS":"S","When":1526413725,"Code":70,"Msg":"CGMiner stats","Description":"cgminer 4.10.0"}],"STATS":[{"CGMiner":"4.10.0","Miner":"1.0.1.3","CompileTime":"Sun Jan 14 00:00:57 CST 2018","Type":"Antminer L3+ Blissz v1.02"}{"STATS":0,"ID":"L30","Elapsed":38187,"Calls":0,"Wait":0.000000,"Max":0.000000,"Min":99999999.000000,"GHS 5s":"579.398","GHS av":577.89,"miner_count":4,"frequency":"431","frequency1":431,"frequency2":431,"frequency3":456,"frequency4":456,"volt1":"13","volt2":"13","volt3":"13","volt4":"13","watt1":243,"watt2":243,"watt3":256,"watt4":256,"fan_num":2,"fan1":3726,"fan2":3780,"temp_num":4,"temp1":60,"temp2":54,"temp3":55,"temp4":53,"temp2_1":67,"temp2_2":62,"temp2_3":64,"temp2_4":61,"temp31":0,"temp32":0,"temp33":0,"temp34":0,"temp4_1":0,"temp4_2":0,"temp4_3":0,"temp4_4":0,"temp_max":60,"Device Hardware%":-0.0000,"no_matching_work":101,"chain_acn1":72,"chain_acn2":72,"chain_acn3":72,"chain_acn4":72,"chain_acs1":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs2":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs3":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs4":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_hw1":36,"chain_hw2":38,"chain_hw3":8,"chain_hw4":19,"chain_rate1":"143.73","chain_rate2":"143.59","chain_rate3":"14

[psycodad]

...

{"STATUS":[{"STATUS":"S","When":1526413725,"Code":70,"Msg":"CGMiner stats","Description":"cgminer 4.10.0"}],"STATS":[{"CGMiner":"4.10.0","Miner":"1.0.1.3","CompileTime":"Sun Jan 14 00:00:57 CST 2018","Type":"Antminer L3+ Blissz v1.02"}{"STATS":0,"ID":"L30","Elapsed":38187,"Calls":0,"Wait":0.000000,"Max":0.000000,"Min":99999999.000000,"GHS 5s":"579.398","GHS av":577.89,"miner_count":4,"frequency":"431","frequency1":431,"frequency2":431,"frequency3":456,"frequency4":456,"volt1":"13","volt2":"13","volt3":"13","volt4":"13","watt1":243,"watt2":243,"watt3":256,"watt4":256,"fan_num":2,"fan1":3726,"fan2":3780,"temp_num":4,"temp1":60,"temp2":54,"temp3":55,"temp4":53,"temp2_1":67,"temp2_2":62,"temp2_3":64,"temp2_4":61,"temp31":0,"temp32":0,"temp33":0,"temp34":0,"temp4_1":0,"temp4_2":0,"temp4_3":0,"temp4_4":0,"temp_max":60,"Device Hardware%":-0.0000,"no_matching_work":101,"chain_acn1":72,"chain_acn2":72,"chain_acn3":72,"chain_acn4":72,"chain_acs1":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs2":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs3":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs4":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_hw1":36,"chain_hw2":38,"chain_hw3":8,"chain_hw4":19,"chain_rate1":"143.73","chain_rate2":"143.59","chain_rate3":"14

Are you sure you pasted everything? That json reply is cut in half and is indeed invalid.

I am not sure if my receive buffer is too large with 32k, I have changed that to 1k and pushed the update to github again. Please make also sure you paste the complete output.

[arnold_mad]

Failed to decode json reply:
Expecting , delimiter: line 1 column 236 (char 235)

{"STATUS":[{"STATUS":"S","When":1526454833,"Code":70,"Msg":"CGMiner stats","Description":"cgminer 4.10.0"}],"STATS":[{"CGMiner":"4.10.0","Miner":"1.0.1.3","CompileTime":"Sun Jan 14 00:00:57 CST 2018","Type":"Antminer L3+ Blissz v1.02"}{"STATS":0,"ID":"L30","Elapsed":79294,"Calls":0,"Wait":0.000000,"Max":0.000000,"Min":99999999.000000,"GHS 5s":"580.898","GHS av":578.18,"miner_count":4,"frequency":"431","frequency1":431,"frequency2":431,"frequency3":456,"frequency4":456,"volt1":"13","volt2":"13","volt3":"13","volt4":"13","watt1":243,"watt2":243,"watt3":256,"watt4":256,"fan_num":2,"fan1":3384,"fan2":3414,"temp_num":4,"temp1":57,"temp2":51,"temp3":53,"temp4":51,"temp2_1":64,"temp2_2":60,"temp2_3":62,"temp2_4":59,"temp31":0,"temp32":0,"temp33":0,"temp34":0,"temp4_1":0,"temp4_2":0,"temp4_3":0,"temp4_4":0,"temp_max":57,"Device Hardware%":-0.0000,"no_matching_work":212,"chain_acn1":72,"chain_acn2":72,"chain_acn3":72,"chain_acn4":72,"chain_acs1":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs2":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs3":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs4":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_hw1":79,"chain_hw2":82,"chain_hw3":13,"chain_hw4":38,"chain_rate1":"147.83","chain_rate2":"144.97","chain_rate3":"142.43","chain_rate4":"145.67"}],"id":1}

[supersonic]

Scrypt starts to work, but it finishes randomly testing - sometimes after 10mins or bit more, even if there is room for undervolt for example with zero HW.
And sometime it does this:

= Running since: 00:10.14, now sleeping for 46.4s =
| 192.168.8.18 [425] |  61  |  60  |  58  |  55  |
+ Current voltages   + 0x80 + 0x60 + 0x80 + 0x80 +
|Errors/min (5min)   | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
|Errors/min (10min)  | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
|Errors/min (15min)  | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
|Errors/min (all)    | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
Chain 1 can be undervolted more (0.00 err/m)
Voltage setting of 0x87 good to test.
Undervolted chain 1 from 0x80 to 0x87
Chain 2 can be undervolted more (0.00 err/m)
Voltage setting of 0x67 good to test.
Undervolted chain 2 from 0x60 to 0x67
Chain 3 can be undervolted more (0.00 err/m)
Voltage setting of 0x87 good to test.
Undervolted chain 3 from 0x80 to 0x87
Chain 4 can be undervolted more (0.00 err/m)
Voltage setting of 0x87 good to test.
Undervolted chain 4 from 0x80 to 0x87
= Running since: 00:12.07, now sleeping for -6.0s =
Traceback (most recent call last):
  File "./l3plus_autotune.py", line 600, in <module>
    time.sleep( REPEAT - (time.time()-now) )
IOError: [Errno 22] Invalid argument

[okg]

what kind of results can one expect? we talking 50watt or 5 ?

Sorry found answer in original thread.

[taserz]

what kind of results can one expect? we talking 50watt or 5 ?

Sorry found answer in original thread.

I am going to ping phil to give it a whirl since I know he has a few L3+. Sadly all miner are in a hosting facility except for one lonely s9 that is pos and been in warranty 5 times. But the OP thread it was 600w vw 800w so that is a 20% saving.

[psycodad]

Scrypt starts to work, but it finishes randomly testing - sometimes after 10mins or bit more, even if there is room for undervolt for example with zero HW.
And sometime it does this:

= Running since: 00:10.14, now sleeping for 46.4s =
| 192.168.8.18 [425] |  61  |  60  |  58  |  55  |
+ Current voltages   + 0x80 + 0x60 + 0x80 + 0x80 +
|Errors/min (5min)   | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
|Errors/min (10min)  | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
|Errors/min (15min)  | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
|Errors/min (all)    | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
Chain 1 can be undervolted more (0.00 err/m)
Voltage setting of 0x87 good to test.
Undervolted chain 1 from 0x80 to 0x87
Chain 2 can be undervolted more (0.00 err/m)
Voltage setting of 0x67 good to test.
Undervolted chain 2 from 0x60 to 0x67
Chain 3 can be undervolted more (0.00 err/m)
Voltage setting of 0x87 good to test.
Undervolted chain 3 from 0x80 to 0x87
Chain 4 can be undervolted more (0.00 err/m)
Voltage setting of 0x87 good to test.
Undervolted chain 4 from 0x80 to 0x87
= Running since: 00:12.07, now sleeping for -6.0s =
Traceback (most recent call last):
  File "./l3plus_autotune.py", line 600, in <module>
    time.sleep( REPEAT - (time.time()-now) )
IOError: [Errno 22] Invalid argument

Thanks for testing the script and reporting back.

Never seen that error here. But I see that we're trying to sleep a negative number of seconds because your ssh connection take longer than 15s each, which makes the computer say "No" in a very Little Brittain like manner.
Will work around this and push an update later today and post here when it is fixed.

I guess that's the reason it never finishes and therefor never really does its job of lowering the voltage.
Have you ever seen it finishing with lines like the following or does it always drop out with an error?

Code:

...
= Running since: 8495s, now sleeping for 24.4s =
Finished tuning, miner stable AFAICS

[supersonic]

Scrypt starts to work, but it finishes randomly testing - sometimes after 10mins or bit more, even if there is room for undervolt for example with zero HW.
And sometime it does this:

= Running since: 00:10.14, now sleeping for 46.4s =
| 192.168.8.18 [425] |  61  |  60  |  58  |  55  |
+ Current voltages   + 0x80 + 0x60 + 0x80 + 0x80 +
|Errors/min (5min)   | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
|Errors/min (10min)  | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
|Errors/min (15min)  | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
|Errors/min (all)    | 0.00 | 0.00 | 0.00 | 0.00 | 0 0 0 0 |
Chain 1 can be undervolted more (0.00 err/m)
Voltage setting of 0x87 good to test.
Undervolted chain 1 from 0x80 to 0x87
Chain 2 can be undervolted more (0.00 err/m)
Voltage setting of 0x67 good to test.
Undervolted chain 2 from 0x60 to 0x67
Chain 3 can be undervolted more (0.00 err/m)
Voltage setting of 0x87 good to test.
Undervolted chain 3 from 0x80 to 0x87
Chain 4 can be undervolted more (0.00 err/m)
Voltage setting of 0x87 good to test.
Undervolted chain 4 from 0x80 to 0x87
= Running since: 00:12.07, now sleeping for -6.0s =
Traceback (most recent call last):
  File "./l3plus_autotune.py", line 600, in <module>
    time.sleep( REPEAT - (time.time()-now) )
IOError: [Errno 22] Invalid argument

Thanks for testing the script and reporting back.

Never seen that error here. But I see that we're trying to sleep a negative number of seconds because your ssh connection take longer than 15s each, which makes the computer say "No" in a very Little Brittain like manner.
Will work around this and push an update later today and post here when it is fixed.

I guess that's the reason it never finishes and therefor never really does its job of lowering the voltage.
Have you ever seen it finishing with lines like the following or does it always drop out with an error?

Code:

...
= Running since: 8495s, now sleeping for 24.4s =
Finished tuning, miner stable AFAICS

Well, yes it finished once or twice after 40mins or so, but there still was zero HW's and way more to go with undervolting. Most of times tho it stops after one pass - about 10-15mins. I just took very old laptop not used in long time, installed ubuntu and pluged it to local network seemed best idea at the time since i got windows on pc

[psycodad]

Failed to decode json reply:
Expecting , delimiter: line 1 column 236 (char 235)

{"STATUS":[{"STATUS":"S","When":1526454833,"Code":70,"Msg":"CGMiner stats","Description":"cgminer 4.10.0"}],"STATS":[{"CGMiner":"4.10.0","Miner":"1.0.1.3","CompileTime":"Sun Jan 14 00:00:57 CST 2018","Type":"Antminer L3+ Blissz v1.02"}{"STATS":0,"ID":"L30","Elapsed":79294,"Calls":0,"Wait":0.000000,"Max":0.000000,"Min":99999999.000000,"GHS 5s":"580.898","GHS av":578.18,"miner_count":4,"frequency":"431","frequency1":431,"frequency2":431,"frequency3":456,"frequency4":456,"volt1":"13","volt2":"13","volt3":"13","volt4":"13","watt1":243,"watt2":243,"watt3":256,"watt4":256,"fan_num":2,"fan1":3384,"fan2":3414,"temp_num":4,"temp1":57,"temp2":51,"temp3":53,"temp4":51,"temp2_1":64,"temp2_2":60,"temp2_3":62,"temp2_4":59,"temp31":0,"temp32":0,"temp33":0,"temp34":0,"temp4_1":0,"temp4_2":0,"temp4_3":0,"temp4_4":0,"temp_max":57,"Device Hardware%":-0.0000,"no_matching_work":212,"chain_acn1":72,"chain_acn2":72,"chain_acn3":72,"chain_acn4":72,"chain_acs1":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs2":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs3":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_acs4":" oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo oooooooo","chain_hw1":79,"chain_hw2":82,"chain_hw3":13,"chain_hw4":38,"chain_rate1":"147.83","chain_rate2":"144.97","chain_rate3":"142.43","chain_rate4":"145.67"}],"id":1}

Thanks, that's something I can chew on. It lacks a , after Type":"Antminer L3+ Blissz v1.02"}. Funny how all versions of cgminer I dealt with output invalid json, though every version places the problem somewhere else.

I have pushed an attempted fix to github, though it's a bit fly-by-night as I don't have a Blissz fw to test it beforehand.

[psycodad]

..

Thanks for testing the script and reporting back.

Never seen that error here. But I see that we're trying to sleep a negative number of seconds because your ssh connection take longer than 15s each, which makes the computer say "No" in a very Little Brittain like manner.
Will work around this and push an update later today and post here when it is fixed.

I guess that's the reason it never finishes and therefor never really does its job of lowering the voltage.
Have you ever seen it finishing with lines like the following or does it always drop out with an error?

Code:

...
= Running since: 8495s, now sleeping for 24.4s =
Finished tuning, miner stable AFAICS

Well, yes it finished once or twice after 40mins or so, but there still was zero HW's and way more to go with undervolting. Most of times tho it stops after one pass - about 10-15mins. I just took very old laptop not used in long time, installed ubuntu and pluged it to local network seemed best idea at the time since i got windows on pc

First, the problem with negative sleep is fixed in the current github version, please update and see if it goes away for you.

Regarding the problem of not getting undervolted enough, could you pm me the contents of one or two report files the tool generates when finished?
I am aware the mechanism for finding the sweet spot voltage-wise is far from perfect yet, but in my tests I always got very accurate results, meaning it improved my manual tuning by a lot.

[supersonic]

..

Thanks for testing the script and reporting back.

Never seen that error here. But I see that we're trying to sleep a negative number of seconds because your ssh connection take longer than 15s each, which makes the computer say "No" in a very Little Brittain like manner.
Will work around this and push an update later today and post here when it is fixed.

I guess that's the reason it never finishes and therefor never really does its job of lowering the voltage.
Have you ever seen it finishing with lines like the following or does it always drop out with an error?

Code:

...
= Running since: 8495s, now sleeping for 24.4s =
Finished tuning, miner stable AFAICS

Well, yes it finished once or twice after 40mins or so, but there still was zero HW's and way more to go with undervolting. Most of times tho it stops after one pass - about 10-15mins. I just took very old laptop not used in long time, installed ubuntu and pluged it to local network seemed best idea at the time since i got windows on pc

First, the problem with negative sleep is fixed in the current github version, please update and see if it goes away for you.

Regarding the problem of not getting undervolted enough, could you pm me the contents of one or two report files the tool generates when finished?
I am aware the mechanism for finding the sweet spot voltage-wise is far from perfect yet, but in my tests I always got very accurate results, meaning it improved my manual tuning by a lot.

Will test it more in couple hours after i get back home and will send You some logs

[arnold_mad]

did you check my full version of the json ?

[psycodad]

did you check my full version of the json ?

I think so, please check this post:

https://bitcointalk.org/index.php?topic=3835841.msg37448663#msg37448663

[philipma1957]

what kind of results can one expect? we talking 50watt or 5 ?

Sorry found answer in original thread.

I am going to ping phil to give it a whirl since I know he has a few L3+. Sadly all miner are in a hosting facility except for one lonely s9 that is pos and been in warranty 5 times. But the OP thread it was 600w vw 800w so that is a 20% saving.

Just got a pm to read this

I have multiple L3+ with the undervolt tool

using just the jstefanop  tool  on freq 359 hash rate of 470 I do 667 watts with minimum volts on all 4 boards. using a corsair ax1500i titanium

this is a savings of over 100 watts.

So my first question is  can this scrypt use a lower volt setting then  0xfe?
here is a 3 board unit with voltage tool

Last login: Tue May  8 10:15:16 on ttys000
philsbeasts-Pro:~ philsbeast$ scp /Users/philsbeast/Downloads/set_voltage root@192.168.0.190:/config
The authenticity of host '192.168.0.190 (192.168.0.190)' can't be established.
RSA key fingerprint is d8:dc:52:5e:0a:c2:63:55:f1:af:3d:ed:7c:4f:e1:ec.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added '192.168.0.190' (RSA) to the list of known hosts.
root@192.168.0.190's password:
set_voltage                                                                     100% 9204     9.0KB/s   00:00    
philsbeasts-Pro:~ philsbeast$ ssh root@192.168.0.190
root@192.168.0.190's password:
root@antMinerTwo2L3:~# cd /config
root@antMinerTwo2L3:/config# ./set_voltage 1fe
Incorrect arguments
Usage:
./set_voltage [chain# 1-4] [voltage in hex]
root@antMinerTwo2L3:/config# ./set_voltage 1 fe

 version = 0x03
reading voltage

 voltage = 0x80
setting voltage
reading voltage

 voltage = 0xfe
Success: Voltage updated!
root@antMinerTwo2L3:/config# ./set_voltage 2 fe

 version = 0x00
Wrong PIC version
root@antMinerTwo2L3:/config# ./set_voltage 3 fe

 version = 0x03
reading voltage

 voltage = 0x80
setting voltage
reading voltage

 voltage = 0xfe
Success: Voltage updated!
root@antMinerTwo2L3:/config#
root@antMinerTwo2L3:/config# ./set_voltage 4 fe

 version = 0x03
reading voltage

 voltage = 0x80
setting voltage
reading voltage

 voltage = 0xfe
Success: Voltage updated!
root@antMinerTwo2L3:/config#




full size
https://i.imgur.com/YNrG63r.png

so after days no Hw's to speak of.

this is manual tuned to lowest volts on the tool and I use freq 359 since it gave me almost no errors

[psycodad]

what kind of results can one expect? we talking 50watt or 5 ?

Sorry found answer in original thread.

I am going to ping phil to give it a whirl since I know he has a few L3+. Sadly all miner are in a hosting facility except for one lonely s9 that is pos and been in warranty 5 times. But the OP thread it was 600w vw 800w so that is a 20% saving.

Just got a pm to read this

I have multiple L3+ with the undervolt tool

using just the jstefanop  tool  on freq 359 hash rate of 470 I do 667 watts with minimum volts on all 4 boards. using a corsair ax1500i titanium

this is a savings of over 100 watts.

So my first question is  can this scrypt use a lower volt setting then  0xfe?
..

Thanks for checking in. No, unfortunately 0xfe equals 254 and ist the highest undervolting you can set.
If you have 4x 0xfe set, there is nothing more you can do to save on energy (except lowering frequency further), this script doesn't help you then as it is about to find the highest undervolting setting where you miners still work reliably.

You could however get possibly more hashes/W if you increase the frequency to the point where you still can run them with 0xfe on all or most blades. And if you are willing to set a frequency where you do not get 0 errors on all chains with 0xfe, my script can help to find the highest possible setting with lowest possible errors.