BITMAIN Antminer S1 support and OverClocking thread

[Bobs Yerunkle]

S1 Overclocked to 400MHz. 204GH/s 12hr avg.
Temperatures at 45/42C. Ambient temperature in S1 area is 24C (wall mount thermostat is two feet below the S1).








Boards flipped around (dogie) and the wifi antenna connector needs to be removed from frame. Fans on both sides.

The fans (www.ebay.com/itm/300981509108) were $5 when I got them. They have raised the price since.
Not high flow but quiet and cheap. I am guessing the fans read 0 due to the chaining. No ill effects yet, it has been a couple days.

Put heat sinks on the inductors (goxed) and I have a bunch of small heat sinks coming and may put some on the chips (just because).

The fan 'brackets' are cut pieces of steel pipe strap (which is useful for a ton of random fastening needs like PSUs, hard drives, hubs, etc)
Feet are resting on thin strips of rubber mat to reduce vibration noise.

The pink zip ties are pink zip ties and the yellow zip ties are yellow zip ties.

This stuff certainly is prod environment material but it entertains me. 

Thanks to:
goxed (https://bitcointalk.org/index.php?topic=348327.0)
John C ([urlhttps://bitcointalk.org/index.php?topic=348327.msg3856863#msg3856863][/url])
dogie (https://bitcointalk.org/index.php?topic=355387.msg3852903#msg3852903).
and this (http://tnerual.eriogerg.free.fr/vimqrc.pdf) because I need to spend some time with one of the most popular *nix text editors.
--------------------------------------------------------------

So how do I actually set up the pool in miner configuration for the 180gh AntMiner?
I don't really have any numbers to enter in there, I normally do stratum+tcp://stratum.mining.eligius.st:3334 (I don't think wzkid likes to give out actual ip numbers)
Then my address for my worker, and x for password.

But, it does not seem to be taking it. The little spinner just sits there and spins, says applying changes to  /etc/config/cgminer     and nothing happens.
I also have a red "unsaved changes = 1" at the to right, no idea what this means....   dhcp.cfg06f37d=domain

I imagine you have this sorted by now but http://stratum.mining.eligius.st:3334 works for eligius with an S1
Click on Unsaved Changes then Save & Apply.

[1714105900]

1714105900

[1714105900]

1714105900

[1714105900]

1714105900

[BadAss.Sx]

Nice setup. I am trying to find out if i can overclock them to 425 or 450Mhz. With your setup i have a better cooling then now. Going to find more of these fans

[klondike_bar]

^very nice idea of turning the boards around! You might consider trying 425MHz if you want to get an extra 10GHS from the unit since you have nice low temps

[Bobs Yerunkle]

^very nice idea of turning the boards around! You might consider trying 425MHz if you want to get an extra 10GHS from the unit since you have nice low temps

Thanks!

The fan on fin idea is from Dogie's Setup thread (https://bitcointalk.org/index.php?topic=355387.msg3852903#msg3852903).

I will probably let it run a bit more and then see what else can be done. So many little projects to do... I still have USB Block Erupters to upgrade to gain 100MH/s...  

Edit: Random detail - The Arctic F12 Pro PWM 120mm fans I'm using are only 57 CFM. You could certainly use stronger fans but may not want to power them from the board.

[mitty]

Does anyone know if it's safe to overclock the new Antminers with the onboard 6-pin PCI-E power connectors when only powering the device with the PCI-E connectors? (nothing in the screw terminals)
I thought the PCI-E power cables felt kind of warm when running on stock clock but it could have just been heat from the unit itself so I want get another opinion.  I'm considering a nice modest 375 overclock.

I guess this question could also be, "is it safe to draw 400-500 W from two 6-pin PCI-E power cables?"

[klondike_bar]

Does anyone know if it's safe to overclock the new Antminers with the onboard 6-pin PCI-E power connectors when only powering the device with the PCI-E connectors? (nothing in the screw terminals)
I thought the PCI-E power cables felt kind of warm when running on stock clock but it could have just been heat from the unit itself so I want get another opinion.  I'm considering a nice modest 375 overclock.

I guess this question could also be, "is it safe to draw 400-500 W from two 6-pin PCI-E power cables?"

It is safe, I would say over 300W per PCI cable is where you run a risk (look at the bitfury threads, where a full rig can draw up to 800W if overclocked, and there are a few cases of burnt-out PCI cables (most of these though were the 2-headed cables, rather than 2 seperate leads).

As long as the cables are good quality and fairly thick, 200-300W each shouldnt be an issue. One of my antminers is powered by a single 4-pin ATX lead (2x12V,2xGND) to each blade with no issue. However, if your cables have zipties or are in a mesh wrap, it could cause a lot of warmth to concentrate in some areas rather than radiating away or being blown away by a fan. If your cables get warm, just direct some airflow over them and they will quickly return to room temps

[mitty]

Does anyone know if it's safe to overclock the new Antminers with the onboard 6-pin PCI-E power connectors when only powering the device with the PCI-E connectors? (nothing in the screw terminals)
I thought the PCI-E power cables felt kind of warm when running on stock clock but it could have just been heat from the unit itself so I want get another opinion.  I'm considering a nice modest 375 overclock.

I guess this question could also be, "is it safe to draw 400-500 W from two 6-pin PCI-E power cables?"

It is safe, I would say over 300W per PCI cable is where you run a risk (look at the bitfury threads, where a full rig can draw up to 800W if overclocked, and there are a few cases of burnt-out PCI cables (most of these though were the 2-headed cables, rather than 2 seperate leads).

As long as the cables are good quality and fairly thick, 200-300W each shouldnt be an issue. One of my antminers is powered by a single 4-pin ATX lead (2x12V,2xGND) to each blade with no issue. However, if your cables have zipties or are in a mesh wrap, it could cause a lot of warmth to concentrate in some areas rather than radiating away or being blown away by a fan. If your cables get warm, just direct some airflow over them and they will quickly return to room temps

Thanks for the info- it seems that some of the cables are just getting warm near the PSU connector but the rest of the cable feels fine.  I'll keep an eye on it but it wasn't warm enough to cause concern.

I'm running two miners on a 1000w PSU and measured ~685w at the wall with no overclock, ~725w at the wall at 375M, and ~780w at the wall at 400M.  I was expecting a lot higher based on the other posts in this topic; maybe I didn't do something right? (I edited /etc/config/asic-freq and power cycled the miners) 
The web interface on most of my Antminers is showing GH/S(avg) to be between 195-203 GH/s and temps are between 42-47. 

[ineedit]

S1 Overclocked to 400MHz. 204GH/s 12hr avg.
...
--------------------------------------------------------------

So how do I actually set up the pool in miner configuration for the 180gh AntMiner?
I don't really have any numbers to enter in there, I normally do stratum+tcp://stratum.mining.eligius.st:3334 (I don't think wzkid likes to give out actual ip numbers)
Then my address for my worker, and x for password.

But, it does not seem to be taking it. The little spinner just sits there and spins, says applying changes to  /etc/config/cgminer     and nothing happens.
I also have a red "unsaved changes = 1" at the to right, no idea what this means....   dhcp.cfg06f37d=domain

I imagine you have this sorted by now but http://stratum.mining.eligius.st:3334 works for eligius with an S1
Click on Unsaved Changes then Save & Apply.

Eligius works with S1's using the normal URL

[Bismarckbkk]

How can I make sure the added fans are running at max capacity?

[ineedit]

How can I make sure the added fans are running at max capacity?

What makes you think that they are not?

[Bobs Yerunkle]

How can I make sure the added fans are running at max capacity?

If you are using PWM fans (https://en.wikipedia.org/wiki/Fan_control#Pulse-width_modulation) then speed should vary depending on cooling needs (as dictated by the controller on the board).
It is hard to know what is happening when chained they way they are since the control panel reads 0. You could test by visual inspecting and listening to the fans at different OC speeds if you wanted to SCIENCE.

If you just want to run the fans at full speed all of the time then you should connect them externally to the PSU per whatever the fan's power needs are (in my case 12v, check fan specs).

RE: inductor heat sinks
One side was attached with heat sink plaster (http://www.ebay.com/itm/360712645612 {there is also a faster US seller for 4x$}) and some heat sinks I already had from 10 years ago.

The other side used these www.amazon.com/gp/product/B005EN6MA0, which are self adhesive. Those were too expensive in my mind but I wanted them quick so I paid the premium.

[mitty]

The "ASIC status" on some of my 400M overclocked Antminers show all x's (i.e., xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx) yet the temps are fine (~44-46) and the hashrate is still showing about 200 GH/s, as well as hashrate on my pool staying the same.  Is this a warning that I'm slowly burning out the unit?

[Bobs Yerunkle]

The "ASIC status" on some of my 400M overclocked Antminers show all x's (i.e., xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx) yet the temps are fine (~44-46) and the hashrate is still showing about 200 GH/s, as well as hashrate on my pool staying the same.  Is this a warning that I'm slowly burning out the unit?

From what I have read "x" means either not enough power to the chip or overheating.

Since the machine is overclocked and pulling more power, perhaps you are on the cusp of the power range of you PSU?
Do you have a Kill-a-watt (or other such device) to measure what the S1 is pulling at the wall?

6. I see "x" in the ASIC Status

For the ASIC status wise, "x" doesn't necessary means the chip is faulty or failing.  If the hashing power seems to be affected, please check your PSU and wiring and reboot and check the status a few times.  It should go back to an average 168-180 gh/s range per AntMiner.

when you are getting "x" on the ASIC chip status, please turn off the antminer and double check your PSU cables are connected well and restart.  Check the status in about 5 minutes and every 30 minutes for a few hours and see if the "x" are showing up constantly at the same chip or randomly changes or completely disappear.

src: https://bitcointalk.org/index.php?topic=344970.0

[ineedit]

The "ASIC status" on some of my 400M overclocked Antminers show all x's (i.e., xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx) yet the temps are fine (~44-46) and the hashrate is still showing about 200 GH/s, as well as hashrate on my pool staying the same.  Is this a warning that I'm slowly burning out the unit?

They can be temporary indicators as well, check with your pool that they are receiving valid shares at the hash rate that your unit is producing.

[warhawk187]

Does anyone know whats the most efficient way of powering these guys? I'm starting to run out of amp circuits (NYC Apt, what can ya do).

Any suggestions for how these should be set up? I have 2 1000W (Gold/Platinum), 1 1050W (Gold), 1 1250W (Gold), with another 2 x 1200W (Gold) and another 1000W (Gold) PSU on the way.

I only have 3 20A circuits and 2 15A for my apartment and I've already had surge protectors/breakers go off infrequently on both and want to make sure I can make sure everything is running well while I'm gone in Mexico for 2 weeks.

[Bismarckbkk]

Sorry to not directly answer your question but id be worried about firehazard. I already had 1 PCI-E connector burned. Maybe find a friend/family or take them with you.

[nwbz]

Many thanks to Bobs Yerunkle, Goxed, Dogie, JohnC and everyone about this Antminer overclocking and cooling tips.

I also overclock my antminer to 400MHz with the result of 200 GHash/S and quite comfortable temperature (40-44).

https://i.imgur.com/meUL7zh.png

Nonetheless, I also tried to overclock to 425MHz, the temperature rose just a little, but came so many HW errors, so I switched back to 400MHz.

PS. Anyone plz suggest the suitable 'minimum difficulty' for the overclocked antminer (200GHash/S)?
Or Antminer would do this for us?
My pool provides 'minimum difficulty' from 2(by default)/4/8/16/64/128/256/512/1024 GHash to choose.

[klondike_bar]

PS. Anyone plz suggest the suitable 'minimum difficulty' for the overclocked antminer (200GHash/S)?
Or Antminer would do this for us?
My pool provides 'minimum difficulty' from 2(by default)/4/8/16/64/128/256/512/1024 GHash to choose.

I would suggest 256. sometimes having a lower difficulty causes too much 'pileup' (at the miner or the pool) and affects hashrate, whereas too high can cause the miner to struggle to complete valid hashes. I use 256 myself at ghash.io and have 197 and 203GH averages from my antminers

[nwbz]

Thank you for your suggestion, klondlike_bar. I am gonna try 256 GHash soon (already tried 32 and 64)

However, as I read from 'Suggestion for how to choose a pool difficulty for miners'
https://bitcointalk.org/index.php?topic=274023.0

ckolivas stated 'Pools should force a minimum difficulty of miner hashrate in GH / 1.8'

I should try some more (128/256) 

PS. Anyone plz suggest the suitable 'minimum difficulty' for the overclocked antminer (200GHash/S)?
Or Antminer would do this for us?
My pool provides 'minimum difficulty' from 2(by default)/4/8/16/64/128/256/512/1024 GHash to choose.

I would suggest 256. sometimes having a lower difficulty causes too much 'pileup' (at the miner or the pool) and affects hashrate, whereas too high can cause the miner to struggle to complete valid hashes. I use 256 myself at ghash.io and have 197 and 203GH averages from my antminers

[radez99]

Nice setup. I am trying to find out if i can overclock them to 425 or 450Mhz. With your setup i have a better cooling then now. Going to find more of these fans

What values should be set for 425 or 450 MHz ?

I mean:
freq_value' 
'chip_freq'  '425'
option 'timeout'