BITMAIN Antminer S1 support and OverClocking thread

[ineedit]

Hi
 do you have any advice how to split AntminerS1 hash power?
 For example 4*50 GHs?
 Is it possible?
thanks

You have edit the /etc/config/cgminer and /etc/init.d/cgminer to either load balance or round robin. I have it working with three pools but I do not have a fourth pool so I have never bothered to try.

[yelloyello]

So a serious question,

What is the difference between 0780 and 4f81 in the overclock settings?

What if I change the frequency to 410?

Anyone tried?

0780 is in decimals 1920
4f81  is in decimals 20353

And what are these options in etc/config/cgminer?

config cgminer 'default'
   option chip_frequency '282'
   option miner_count '24'

[jgm_coin]

I'm selling my antminer on ebay:

http://r.ebay.com/6f6Y3a

Already overclocked to 192 GH/s for you.  I will accept BTC for cheaper.  Send me PM

[machinationus]

I have my 2 antminers running with 1000watt nzxt psu and they do fine however i do not use high overclocks if you use 400 mhz or higher 500 tot 525 watt per device is needed minimal
Asic settings found on different topics:

mhz : 387.5
freq_value '5e85'
timeout 38

mhz :393.75
freq_value '5f05'
timeout 38

One miner still constant turns all the O into X after several hours of mining, i have no clue why this constant happens
The Hashrate seems to be according the specified range, but i am worried about this one

Power usage from the 387 setting seems to be 490watt

I got the following, with a i600 coolermaster

412 watt 191 GH/s 4e81 timeout 38  "375"  decimal 20097  ~1 hour HW=193

(http://www.mathsisfun.com/binary-decimal-hexadecimal-converter.html)

435 watt 192 GH/s 4f81 timeout 35 "400"   decimal 20353  ~1 hour HW=11,623



we need to know the relationship between the timeout and speed ??

[MrTeal]

I'm selling my antminer on ebay:

http://r.ebay.com/6f6Y3a

Already overclocked to 192 GH/s for you.  I will accept BTC for cheaper.  Send me PM

LOL... How much cheaper? Bitmain is supposed to start shipping again tomorrow and while it will likely take a few days to get through the backlog, they're only charging ~$1050 at the current price of BTC for an S1.

[air139]

This thread is for support related discussions of the BITMAIN Antminer S2 hashing device

How to overclock your miner to 400MHz / 205GH/s. My suggestion is to clock them at 375. 400 is a bit too harsh on the voltage regulator.
All clock speeds including 375MHz work on both revision of boards.

The miner will draw 500 - 525Watts at the wall (110V mains) after this setting

A) ssh into your miner

B) edit the following file
/etc/config/asic-freq

C) change timeout to 36, freq_value to '0780', 'chip_freq' to 400.
For miners with build date 2013/11/4

Code:

root@antMiner2:/etc/config# vim asic-freq 

package 'cgminer'

config 'asic-freq' 'default'
        #option 'freq_value'      '0780'        #400M

        #option 'freq_value'      '4e81'        #375M
        #option 'chip_freq'     '375'
        #option 'timeout'       '37'

        #option 'freq_value'    '4d81'  #350M
        #option 'chip_freq'     '350'
        option 'timeout'        '36'
        #option 'freq_value'    '4c81'  #325M
        #option 'chip_freq'     '325'
        option 'freq_value'     '0780'  #350M
        option 'chip_freq'      '400'
        #option 'freq_value'    '0580'  #300M
        #option 'chip_freq'     '300'
        #option 'freq_value'    '0981'  #250M
        #option 'chip_freq'     '250'
        #option 'freq_value'    '4f02'  #190M
        #option 'chip_freq'     '190'

For miners with build date 2013/11/18

Code:

package 'cgminer'

config 'asic-freq' 'default'
        option 'freq_value'      '4f81'        #400M
        option 'chip_freq'     '400'
        option 'timeout'       '35'

        #option 'freq_value'      '4e81'        #375M
        #option 'chip_freq'     '375'
        #option 'timeout'       '37'

        #option 'freq_value'     '4d81'  #350M
        #option 'chip_freq'      '350'
        #option 'timeout'        '40'

        #option 'freq_value'    '4c81'  #325M
        #option 'chip_freq'     '325'
        #option 'timeout'       '40'

        #option 'freq_value'    '0b81'  #300M
        #option 'chip_freq'     '300'
        #option 'timeout'       '45'

        #option 'freq_value'    '0981'  #250M
        #option 'chip_freq'     '250'
        #option 'timeout'       '55'

        #option 'freq_value'    '0781'  #200M
        #option 'chip_freq'     '200'
        #option 'timeout'       '70'

        #option 'freq_value'    '4f02'  #193M
        #option 'chip_freq'     '193'
        #option 'timeout'       '70'

How to RESET your OpenWRT card in case of an unresponsive miner

Please follow the steps in the instruction manual provided by BITMAIN below.
https://github.com/AntMiner/AntGen1/blob/master/manual/AntMiner%20FAQ.pdf?raw=true
https://github.com/AntMiner/AntGen1/tree/master/manual


Please donate if it works for you: 1FrW5ka9CRVWdL9EM4Jut9eHWU1a1biyy5

How stable are they? What chance is there that they will die suddenly and I will regret it?

[Darthswan]

Question about the timeout number.  I have notice a range from 36 to 38.  What is the purpose of the timeout number?  What is its effect on the miner.  Just curious.


With the right power supply and proper cooling, they are pretty stable 375.  I have 2 running at 387.

[Drai]

Good Evening.

I seem to have one x of my Ant ASIC status.

Should I be concerned? Currently mining at 375. PSU Seasonic 650w Gold.

Thanks for any feedback.

I have xes in ASIC Status at 400mhz, at 43-44°C with
elapsed 3d 3h 52m 14s
 92,223 accepted, 245 rejected

Should I be concerned?

[Gamer67]

No.

As long as it is still hashing and accepting shares on your pool then there is nothing to worry about.

If you do not like seeing the xxx just restart and it should clear.

[vvhiteknight]

Hey, just thought id remind all you guys that are getting your ants tomorrow to check out my setup and over clocking videos which will get you setup and over clocked inside 15 minutes.

Setup Video: https://www.youtube.com/watch?v=akw9GAeIH5k

Over clocking Video: https://www.youtube.com/watch?v=72FU4Z_n0B4

[Darthswan]

I went throughout the entire thread, is their a consensus on an acceptable HWI% for the different speeds

400
393.75
387.25
375
350

I'm curious.  I have clocked 2 of them up to 393M.

Thanks.

[Drai]

I went throughout the entire thread, is their a consensus on an acceptable HWI% for the different speeds

400
393.75
387.25
375
350

I'm curious.  I have clocked 2 of them up to 393M.

Thanks.

[freebit13]

I went throughout the entire thread, is their a consensus on an acceptable HWI% for the different speeds

400
393.75
387.25
375
350

I'm curious.  I have clocked 2 of them up to 393M.

Thanks.

I'm not sure on the consensus, but 1% or less is acceptable... more than that and you probably need more cooling or more power

[samsonn25]

There is a special formula for that not just the raw HW number.

[klondike_bar]

There is a special formula for that not just the raw HW number.

^this. IIRC its     error % = HW / (HW+diffR+DiffA) x 100%

most of my units run between 0.15-1.05% error rates. I would say that up to 2% is fine, but more then that means you should tune it down slightly and it will be beneficial.

(ie: 2% of 204GH at 400MHz is 4.08GH lost to errors. if that can be reduced to 0.5% errors by tuning down to 393 (~200GH) or 387.5 (197GH) then you will have almost the same hashrate but a little less fan noise and electricity draw to achieve it

[ineedit]

There is a special formula for that not just the raw HW number.

^this. IIRC its     error % = HW / (HW+diffR+DiffA) x 100%

most of my units run between 0.15-1.05% error rates. I would say that up to 2% is fine, but more then that means you should tune it down slightly and it will be beneficial.

(ie: 2% of 204GH at 400MHz is 4.08GH lost to errors. if that can be reduced to 0.5% errors by tuning down to 393 (~200GH) or 387.5 (197GH) then you will have almost the same hashrate but a little less fan noise and electricity draw to achieve it

Nearly, it is HW / (HW + DiffA + DiffR + DiffS) = %

[ineedit]

A couple of people have sent me pm's asking how to load balance pools on an Antminer so I though I would put it here

Change the /etc/init.d/cgminer file from

        PARAMS="$AOPTIONS $POOL1 $POOL2 $POOL3 $_pb --api-allow $_aa --api-listen"
to
        PARAMS="$AOPTIONS $POOL1 $POOL2 $POOL3 $_pb $_mo --api-allow $_aa --api-listen"


change the /etc/config/cgminer from

        option more_options '--quiet'
to
        option more_options '--quiet --load-balance'       

The shares will then be evenly split across the pools that you have in your pool list. This works and I have been using this for a while.

You should also be able to adjust the quota or weighting of the pools by putting a quota in front of each pool in the pool list. I have not bothered to change the weighting as I have had no need but just enter the number followed by a ;  on the miner config page, for example

pool1          60;http://poola:port
pool2          40;http://poolb:port

and you will get a 60/40 split.

[MrTeal]

(ie: 2% of 204GH at 400MHz is 4.08GH lost to errors. if that can be reduced to 0.5% errors by tuning down to 393 (~200GH) or 387.5 (197GH) then you will have almost the same hashrate but a little less fan noise and electricity draw to achieve it

Not necessarily. A HW error is just a returned share that for whatever reason doesn't check out when returned from the mining device to the mining software. Let's say you're scanning a 32 bit nonce range and there are 5 valid diff 1 shares in it. You could have a situation where one of those shares was found but got malformed when it was returned, and the mining software would reject it. You'd then have 1 HW error and 4 Diff1 shares. You could however also have a situation where all 5 diff1 shares are returned correctly, but it also returned an extra share or two that are not actually valid shares. In that case, you'd find all 5 of the valid nonces and just have one extra HW error. You didn't actually lose any hashrate to it.

[erickjaz]

On a different note:

Now that the new firmware for the S1 is out with CGminer 3.12 should we be able to proxy several S1's?  I tried it to no avail, though I am not a CGminer expert.

E

[Darthswan]

There is a special formula for that not just the raw HW number.

^this. IIRC its     error % = HW / (HW+diffR+DiffA) x 100%

most of my units run between 0.15-1.05% error rates. I would say that up to 2% is fine, but more then that means you should tune it down slightly and it will be beneficial.

(ie: 2% of 204GH at 400MHz is 4.08GH lost to errors. if that can be reduced to 0.5% errors by tuning down to 393 (~200GH) or 387.5 (197GH) then you will have almost the same hashrate but a little less fan noise and electricity draw to achieve it

Thanks for the info:

I have 2 clocked at 393 and the other 1 at 387.  My HWI% is ranging from 1.32%, 1.65%, and 1.51%.  The power supply should be fine, I have a HX750.  So i'm going to add more fans to it, and see if I can get that lower.

On the same note, has anyone found a way to increase the speed of the fan without cutting the blue wire? 

Thanks again.