ANTMINER S3+ Discussion and Support Thread

[nexus99]

The settings were provided by bitmain in the .conf file.

[philipma1957]

Questions if you don't mind:

1.  What is your wattage at the wall for 231.25M (please include other relevant data such as ambient room temp and PSU efficiency).
2.  What are the rest of the parameters (i.e. 'freq_value' and 'timeout') for 231.25M?
3.  How does one come up with the appropriate 'freq_value' and 'timeout' for a particular frequency?  Is there a certain formula/equation involved or is it pretty much a trial and error method?

Pardon my ignorance in this matter.

sure here is a full config file

with bonus extras

bold type are ones I found and tested

timeout is found by using the close timeout on the factory set ones.

 an example of a 'timeout' for 256.25 was done at 15   I found it to work  I tried it because factory freq 250 uses freq 16


root@antMiner:~# vi /etc/config/asic-freq

i...package 'cgminer'

config 'asic-freq' 'default'
        

      #option  'freq_value'   '1406'     #Freq at 256.25M = 512 gh uses  410 watts
       #option  'chip_freq'     '256.25'
       #option  'timeout'        '15'

        #option 'freq_value'    '0982'  # Freq at 250M = 500gh uses 400 watts
        #option 'chip_freq'     '250'
        #option 'timeout'       '16'

       #option  'freq_value'    '1306' # Freq at 243.75M = 490gh uses 390watts
        #option  'chip_freq'     '243.75'
        #option  'timeout'       '16'

        option 'freq_value'    '1286'  # Freq at 237.5M = 480gh uses 380 watts
        option 'chip_freq'     '237.5'
        option 'timeout'       '17'

        #option  'freq_value'    '1206'   # Freq at 231.25M = 465gh uses 375 watts
        #option  'chip_freq'     '231.25'
        #option   'timeout'       '17'



        #option 'freq_value'    '0882'  #225M
        #option 'chip_freq'     '225'
        #option 'timeout'       '18'

        #option 'freq_value'    '1106'  #218.75M
        #option 'chip_freq'     '218.75'
        #option 'timeout'       '18'

        #option 'freq_value'    '1086'  #212.5M
        #option 'chip_freq'     '212.5'
        #option 'timeout'       '18'

        #option 'freq_value'    '1006'  #206.25M
        #option 'chip_freq'     '206.25'
        #option 'timeout'       '19'

        #option 'freq_value'    '0782'  #200M
        #option 'chip_freq'     '200'
        #option 'timeout'       '20'

        #option 'freq_value'    '1f07'  #196M
        #option 'chip_freq'     '196'
        #option 'timeout'       '20'

        #option 'freq_value'    '0f03'  #193M
        #option 'chip_freq'     '193'
        #option 'timeout'       '20'

        #option 'freq_value'    '0d83'  #175M
        #option 'chip_freq'     '175'
        #option 'timeout'       '23'

        #option 'freq_value'    '0b83'  #150M
        #option 'chip_freq'     '150'
        #option 'timeout'       '27'

        #option 'freq_value'    '0983'  #125M
        #option 'chip_freq'     '125'
        #option 'timeout'       '33'

        #option 'freq_value'    '0783'  #100M
        #option 'chip_freq'     '100'
        #option 'timeout'       '40'

[7queue]

Is this the source for what Bitmain is using to roll the cgminer provided on the S3

https://github.com/bitmaintech/cgminer

8 )

Last commit to the code there:
LingchaoXu authored on 31 Dec 2013

If the S3 upgrade re-uses the controller from the S1, it does from the looks of it, would folding the customizations from their github repository into yours work?

8 )

[IYFTech]

With a lot of polling from a big customer they eventually released the S2 driver.

The S3 driver we've had no luck getting out of them.

Well, if a big customer can get them to release the code - I'm pretty sure that the community here can manage it. By my reckoning, if only 50% of cgminer users kicked up a fuss, they'd be completely swamped - same goes for S3 users.

Come on everyone - let's get it done!!! 

[PlanetCrypto]

Does anyone know how Kano was able to write his own version for the S1? Does he have supporting documents? Or did he just come up with a work around?

They had released the code for the driver for the S1.

With a lot of polling from a big customer they eventually released the S2 driver.

The S3 driver we've had no luck getting out of them.

To whom it may concern:

We WERE loyal BITMAIN customers owing to their shipping model, quality of product, and expeditious order fulfillment.
We WERE gearing up to place a large S3+ order.
That capex expenditure is now on hold.
We are now investigating a large purchase of BTC Garden AM-V2's instead.
2 of which are under going testing as I type this and, so far, we are impressed.

BITMAIN's illegal refusal to release current driver code for their S3 product line places the viability of mining with S3's in jeopardy.
BITMAIN's current "version" of the S3 code "has more bugs than a bait store".
This is exemplified by the activity in this thread.
The S3's are not the profitable "plug-n-play" S1 miners we came to so dearly love.

Additionally, we support the tireless, largely uncompensated, professional work done by CK and Kano.
We intend on voicing our discontent with BUTMAIN's pre-order business model (and yes we got taken by BFL), illegal refusal to release code as mandated by the GPL License, and buggy S/W by exercising our purchasing dollars/BTC elsewhere.

Wake up BITMAIN "you're not the only game in town".

PlanetCrypto

[BITMAIN]

Dear Community,

Our truly apologies for the mis-understanding reported in this Support Thread. We are very willing to share the cgminer code to the Community.
Just reviewed our internal resource again, there is a little mis-communicaton between Customer Service Team and R&D Department. We thought this task was done 3 weeks ago..  
However, R&D Department is repacking the cgminer code and will upload it to GitHub.com in 3 hours.

Any advice from the Community and Global Customers is appreciated, feel free to contact us via PM or info@bitmaintech.com.

Thank you!

[squashpile]

Ha.. well there ya go  
Call off the raid.

[hurricandave]

Dear Community,

Our truly apologies for the mis-understanding reported in this Support Thread. We are very willing to share the cgminer code to the Community.
Just reviewed our internal resource again, there is a little mis-communicaton between Customer Service Team and R&D Department. We thought this task was done 3 weeks ago..  
However, R&D Department is repacking the cgminer code and will upload it to GitHub.com in 3 hours.

Any advice from the Community and Global Customers is appreciated, feel free to contact us via PM or info@bitmaintech.com.

Thank you!

And what about support documents to BFG Miner for Development?

[sirux]

Dear Community,

Our truly apologies for the mis-understanding reported in this Support Thread. We are very willing to share the cgminer code to the Community.
Just reviewed our internal resource again, there is a little mis-communicaton between Customer Service Team and R&D Department. We thought this task was done 3 weeks ago..  
However, R&D Department is repacking the cgminer code and will upload it to GitHub.com in 3 hours.

Any advice from the Community and Global Customers is appreciated, feel free to contact us via PM or info@bitmaintech.com.

Thank you!

Great news, thank-you!

[nwoolls]

Dear Community,

Our truly apologies for the mis-understanding reported in this Support Thread. We are very willing to share the cgminer code to the Community.
Just reviewed our internal resource again, there is a little mis-communicaton between Customer Service Team and R&D Department. We thought this task was done 3 weeks ago..  
However, R&D Department is repacking the cgminer code and will upload it to GitHub.com in 3 hours.

Any advice from the Community and Global Customers is appreciated, feel free to contact us via PM or info@bitmaintech.com.

Thank you!

Great news! Will there be documentation as well?

[PlanetCrypto]

Dear Community,

Our truly apologies for the mis-understanding reported in this Support Thread. We are very willing to share the cgminer code to the Community.
Just reviewed our internal resource again, there is a little mis-communicaton between Customer Service Team and R&D Department. We thought this task was done 3 weeks ago..  
However, R&D Department is repacking the cgminer code and will upload it to GitHub.com in 3 hours.

Any advice from the Community and Global Customers is appreciated, feel free to contact us via PM or info@bitmaintech.com.

Thank you!

"Fool me once, foo on you. Fool me twice, foo on me".

When CK posts that he has access to the code and it's what runs in the S3 then I'm a believer.

[edonkey]

Dear Community,

Our truly apologies for the mis-understanding reported in this Support Thread. We are very willing to share the cgminer code to the Community.
Just reviewed our internal resource again, there is a little mis-communicaton between Customer Service Team and R&D Department. We thought this task was done 3 weeks ago..  
However, R&D Department is repacking the cgminer code and will upload it to GitHub.com in 3 hours.

Any advice from the Community and Global Customers is appreciated, feel free to contact us via PM or info@bitmaintech.com.

Thank you!

Thanks for responding to the community. We all look forward to the posted code.

But next time you release a product, can you make sure that posting the source is part of your release plan? Do it early and not as an afterthought. You could save all of us a great deal of wasted time and anxiety.

[eoakland]

Bitmain, what about my RMA...i have been dealing with someone named "Tim"...who requested/forced me to purchase a multimeter device to perform tests...then he never responds with information about my RMA.  he says they have to look for courier tracking information to prove my s3 is from your factory.  And this is after i gave him the order ID.

I am sitting on an expensive paperweight...2 weeks already.  Can i please get a muthafucking RMA

[sirux]

Bitmain, what about my RMA...i have been dealing with someone named "Tim"...who requested/forced me to purchase a multimeter device to perform tests...then he never responds with information about my RMA.  he says they have to look for courier tracking information to prove my s3 is from your factory.  And this is after i gave him the order ID.

I am sitting on an expensive paperweight...2 weeks already.  Can i please get a muthafucking RMA

Wow thats sad

[BITMAIN]

Bitmain, what about my RMA...i have been dealing with someone named "Tim"...who requested/forced me to purchase a multimeter device to perform tests...then he never responds with information about my RMA.  he says they have to look for courier tracking information to prove my s3 is from your factory.  And this is after i gave him the order ID.

I am sitting on an expensive paperweight...2 weeks already.  Can i please get a muthafucking RMA

Dear eoakland,

May you PM your email box?

[eoakland]

Bitmain, what about my RMA...i have been dealing with someone named "Tim"...who requested/forced me to purchase a multimeter device to perform tests...then he never responds with information about my RMA.  he says they have to look for courier tracking information to prove my s3 is from your factory.  And this is after i gave him the order ID.

I am sitting on an expensive paperweight...2 weeks already.  Can i please get a muthafucking RMA

Dear eoakland,

May you PM your email box?

PM sent.

[Master One]

So if cgminer in the S3 is based on an older (buggy?) version and not the current one from ckolivas, and Bitmaintech indeed is now releasing the source code of their version, who is going to update the S3 firmware then?

After all Bitmaintech has (had) access to the source code of ckolivas' current version all along, so that issue never was depending on the release of their patches and/or driver.

[visdude]

The settings were provided by bitmain in the .conf file.

If you meant the /etc/config/asic-freq file, 231.25M is not in it which obviously lies in between 237.5M and 225M, hence my questions about it's source or how it is derived.

[philipma1957]

The settings were provided by bitmain in the .conf file.

If you meant the /etc/config/asic-freq file, 231.25M is not in it which obviously lies in between 237.5M and 225M, hence my questions about it's source or how it is derived.

I told you I figured it out by studying the available patterns in the oem  guide.


the jumps provided are 6.25  or 12.50

so a chart was made

256.25
250.00
243.75
237.5
231.25
225
218.75
212.5

the key to figure this out is to see how the hex code moves  with the provided freqs  then  try what should work in the newer numbers.

the hex code for 212.5 is  1086  provided

the hex code for 218.75 is 1106      so a jump of 6.25    jumped code  0020

so i looked for a freq  that ended in 86  and 237.5 the code   was  1286       so add 6.25 get 243.75     jump the 1286 to 1306

  so I figured the 243.75 as 1306   the timeout was  a guess as to use 16 or 17.

 this is how I got  

 freq 243.75  and code 1306 and time out 16   now I had 1 extra working config


now look at more patterns





       #option  'freq_value'   '1406'     #Freq at 256.25M = 512 gh uses  410 watts
       #option  'chip_freq'     '256.25'
       #option  'timeout'        '15'

       #option 'freq_value'    '0982'  # Freq at 250M = 500gh uses 400 watts
        #option 'chip_freq'     '250'
        #option 'timeout'       '16'

       #option  'freq_value'    '1306' # Freq at 243.75M = 490gh uses 390watts
        #option  'chip_freq'     '243.75'
        #option  'timeout'       '16'

        option 'freq_value'    '1286'  # Freq at 237.5M = 480gh uses 380 watts
        option 'chip_freq'     '237.5'
        option 'timeout'       '17'

        #option  'freq_value'    '1206'   # Freq at 231.25M = 465gh uses 375 watts
        #option  'chip_freq'     '231.25'
        #option   'timeout'       '17'



        #option 'freq_value'    '0882'  #225M
        #option 'chip_freq'     '225'
        #option 'timeout'       '18'

        #option 'freq_value'    '1106'  #218.75M
        #option 'chip_freq'     '218.75'
        #option 'timeout'       '18'

        #option 'freq_value'    '1086'  #212.5M
        #option 'chip_freq'     '212.5'
        #option 'timeout'       '18'

        #option 'freq_value'    '1006'  #206.25M
        #option 'chip_freq'     '206.25'
        #option 'timeout'       '19'

        #option 'freq_value'    '0782'  #200M
        #option 'chip_freq'     '200'
        #option 'timeout'       '20'

        #option 'freq_value'    '1f07'  #196M
        #option 'chip_freq'     '196'
        #option 'timeout'       '20'

        #option 'freq_value'    '0f03'  #193M
        #option 'chip_freq'     '193'
        #option 'timeout'       '20'

        #option 'freq_value'    '0d83'  #175M
        #option 'chip_freq'     '175'
        #option 'timeout'       '23'

        #option 'freq_value'    '0b83'  #150M
        #option 'chip_freq'     '150'
        #option 'timeout'       '27'

        #option 'freq_value'    '0983'  #125M
        #option 'chip_freq'     '125'
        #option 'timeout'       '33'

        #option 'freq_value'    '0783'  #100M
        #option 'chip_freq'     '100'
        #option 'timeout'       '40'




200 = 0782

225 = 0882

250 = 0982




243.75 = 1306

231.25 = ?        so  these are all counting systems   the patterns must repeat.  the freq is base 10  the code is base 16    so   what is that question mark  1206  tried it with timeout of 17  bingo worked

218.75 =1106


once

218.75 = 1106
231.25 = 1206
243.75 = 1306



finding   256.25 = 1406   was easy  used timeout of 15 since 250 uses 16   

 feel free to play with the time out numbers  they may be off by 1

the code in hex is correct for

256.25
250
243.75
237.5
231.25
225
218.75
212.5
206.25
200.00

I did not test lower then 200.

[suchmoon]

So if cgminer in the S3 is based on an older (buggy?) version and not the current one from ckolivas, and Bitmaintech indeed is now releasing the source code of their version, who is going to update the S3 firmware then?

After all Bitmaintech has (had) access to the source code of ckolivas' current version all along, so that issue never was depending on the release of their patches and/or driver.

Having access is not everything that's needed, there's knowledge/experience and development time and QA effort etc. Having the driver out in the open will hopefully expedite the process.