ANTMINER S3+ Discussion and Support Thread

[dogie]

Let me explain as you can see from the images below the S1 has a best share of 5.5 million, but the S3 is still sitting around the 1k? WTF???

So does this mean that the S1 is truly better then the S3 even though the S3 have 4-5 times the hashrate or is there actually something wrong with the firmware yet again. IDK if anyone has this same problem but me myself I think that S3 should be around 20-30 million by now after 12 hours of runtime. I don't believe this is "luck" as it is on a pool and yes have tried others with the same results.

I have posted this issue before on a different thread and someone told me it was a "luck" issue and if so then that means the S1 is FAR better then the S3 anyday and I don't believe that believe this is a firmware or maybe CGminer software related issue. Because I still don't think the S1 should have such a high best share reported and the S3 be so low and no the miner does not act strange in anyway. It runs fine and is on a EVGA 750 B2 PSU (brand new) also have tried it on the CX 500 PSU the S1 is on and same problem.

Also ran both of them for 3 days on p2pool and the S1 had a BS of 14 million + but the S3 didn't even hit 8 million so therefore the S1 was finding all the shares and the S3 didn't do a thing.

Best share doesn't really mean anything when you're pool mining, as you're getting paid to attempt to find a block, not when you find a block. Even then, best share is just a representation of the luck a miner receives.

In this case as well, best share's coding has been a bit weird on the last few miners so it shouldn't be looked at in any detail. I'm surprised you're seeing any number in there at all when you're difficulty is 2^n [which usually makes it display 0].

[1715519167]

1715519167

[dogie]

Dogie, you have a copy of the 10/24 firmware I can download?

https://bitcointalk.org/index.php?topic=883197.msg10170014#msg10170014

1219 (the fixed version of 1024) is below

Download link is here. Rolls back to the last good patch in Oct (1024), but fixes voltage adjustment and the important reset button. MD5: 7DF1B12D28B25E10AAC37E2EAFC7397B

[This is 1219 = 19th Dec 2014 firmware]

And 1024 is rehosted on mega here. Couldn't find the MD5 though.

Are both of these from BITMAIN?

I have been running some on 10-24 and 12-19. The better overclockers I run on the strongest PSUs since the voltage does have an effect. The poorer performers I have been running 12-19 with until I get to whichever unit and figure out why it is hashing slower than the others.

In my experience it has been the unit which decides which firmware is better. Or there is a possibility I'm not doing something right.
I set the poor performers to 218 / 0720

12-19 and 10-24 should be identical performance wise, but 12-19 includes the extremely important reset button fix. Please upgrade to that, you'll be grateful when a unit soft bricks and you don't have to send it in for repair.

[dogie]

Thanks! Did you got any replay?

No reply unfortunately. I'm sure this has been asked about the S1 and someone came up with an ebay listing but with forum search down I was unable to locate it. I've found listings on globalsources which just refer to similar looking cables as X pin flat ribbon cables, that's the closest I can get.

Same on my end. In most cases I get fast respond but not this time... S1 cable will not work. It is different... I looked...

I think a lot of people were travelling this week, so hopefully back to normal and some replies on Monday.

[pekatete]

12-19 and 10-24 should be identical performance wise, but 12-19 includes the extremely important reset button fix.

Identical is no where near accurate .... have you any guidance as to what versions of cgminer ship with either?

[aarons6]

Maybe someone here can explain these strange issues I am having between an antminer S1 (works like it should) and antminer S3 (doesn't work right to me)...

 This is my problem tried all firmware versions and have found the same issues and I recognize it as a problem because my S3 is not working as it should. Let me explain as you can see from the images below the S1 has a best share of 5.5 million, but the S3 is still sitting around the 1k? WTF???

 So does this mean that the S1 is truly better then the S3 even though the S3 have 4-5 times the hashrate or is there actually something wrong with the firmware yet again. IDK if anyone has this same problem but me myself I think that S3 should be around 20-30 million by now after 12 hours of runtime. I don't believe this is "luck" as it is on a pool and yes have tried others with the same results.

Antminer S1



Antminer S3




I have posted this issue before on a different thread and someone told me it was a "luck" issue and if so then that means the S1 is FAR better then the S3 anyday and I don't believe that believe this is a firmware or maybe CGminer software related issue. Because I still don't think the S1 should have such a high best share reported and the S3 be so low and no the miner does not act strange in anyway. It runs fine and is on a EVGA 750 B2 PSU (brand new) also have tried it on the CX 500 PSU the S1 is on and same problem.

Also ran both of them for 3 days on p2pool and the S1 had a BS of 14 million + but the S3 didn't even hit 8 million so therefore the S1 was finding all the shares and the S3 didn't do a thing.

your s3 has a TON of hwe.. it might be worth a repaste.
plus i dont see your difficulty..

update to the 12/19 firmware and see how that works..

if the difficulty is low on the s3 the shares will be low..

[sloopy]

I know this has previously been discussed but the lack of search is killing my time looking for a refresher. I appreciate your patience when explaining these things. If you know of a page in this thread or another thread where I can find these answers please point me in the right direction.

All of my discussion is 'US residential" 240vAC (2 wires each measuring 120vAC and a ground) and not what I normally see in Industrial locations 240 / 480 3 phase. The same which runs most dryers in your home.

 I am moving some miners over to a new 240vAC circuit and before actually pulling the plug so to speak I wanted to re check my numbers.
I remember being told and I think it was this thread that each blade has the potential to pull 13.8 amps. When I start running my own numbers based off the product specifications at BITMAIN I see 366 watts 'at the wall' being specified.

Re-reading page 1 of this thread does indeed quote both numbers. This leaves me confused.
I assume the 13.83 number is quoted at 120vAC. Which if I run 366w at 120vAC I expect 3.05 AMPS.
At 240vAC and 366w I expect 1.525a and so on.

Please help me understand this point. I do have a licensed electrician who approves my work, but, I must give him the correct information for his approval to be worth something.

I am wondering if this is the difference between a quoted nominal voltage - 240vac 1.525 Amps and the number thrown out of 13.83 at 120vAC being 1659.6 watts and respectively 240vAC at 6.915 Amps being the same wattage as only amperage draw would be affected with the voltage change.

Is the quoted 366 watts a nominal draw where 1660 watts is similar to the inrush from a motor starting and that is what the S3 or S3+ would draw at Max load during startup?

Some breakers have inrush protection to handle the additional load / draw at power up. A timer which can be set, usually a dial you turn to adjust the amount of time the inrush is allowed. Is this type of thing the difference in the quoted numbers or is it something else I am missing entirely?

I appreciate your help and hope anyone reading this knows I am doing everything possible to be safe and so should you. I double, triple, and quadruple check my breaker type and size, wire size and length, psu specifications and ratings, the wire size going to the miner and connection points, ie the connectors themselves where the wire attaches to the miner or any other components.
Noise levels are not a concern to me at this time. I love the noise. It blocks out other distractions.

Thanks for your time.

[pekatete]

^^^ You are mixing up numbers, the amperage quoted is at 12v DC (twelve volts direct current) but neither 120v AC nor 240v AC.

[dogie]

I know this has previously been discussed but the lack of search is killing my time looking for a refresher. I appreciate your patience when explaining these things. If you know of a page in this thread or another thread where I can find these answers please point me in the right direction.

All of my discussion is 'US residential" 240vAC (2 wires each measuring 120vAC and a ground) and not what I normally see in Industrial locations 240 / 480 3 phase. The same which runs most dryers in your home.

 I am moving some miners over to a new 240vAC circuit and before actually pulling the plug so to speak I wanted to re check my numbers.
I remember being told and I think it was this thread that each blade has the potential to pull 13.8 amps. When I start running my own numbers based off the product specifications at BITMAIN I see 366 watts 'at the wall' being specified.

Re-reading page 1 of this thread does indeed quote both numbers. This leaves me confused.
I assume the 13.83 number is quoted at 120vAC. Which if I run 366w at 120vAC I expect 3.05 AMPS.
At 240vAC and 366w I expect 1.525a and so on.

Please help me understand this point. I do have a licensed electrician who approves my work, but, I must give him the correct information for his approval to be worth something.

I am wondering if this is the difference between a quoted nominal voltage - 240vac 1.525 Amps and the number thrown out of 13.83 at 120vAC being 1659.6 watts and respectively 240vAC at 6.915 Amps being the same wattage as only amperage draw would be affected with the voltage change.

Is the quoted 366 watts a nominal draw where 1660 watts is similar to the inrush from a motor starting and that is what the S3 or S3+ would draw at Max load during startup?

Some breakers have inrush protection to handle the additional load / draw at power up. A timer which can be set, usually a dial you turn to adjust the amount of time the inrush is allowed. Is this type of thing the difference in the quoted numbers or is it something else I am missing entirely?

I appreciate your help and hope anyone reading this knows I am doing everything possible to be safe and so should you. I double, triple, and quadruple check my breaker type and size, wire size and length, psu specifications and ratings, the wire size going to the miner and connection points, ie the connectors themselves where the wire attaches to the miner or any other components.
Noise levels are not a concern to me at this time. I love the noise. It blocks out other distractions.

Thanks for your time.

super tldr, the referred 13.8A will be at 12V (twelve), from the PSU = 165W. Its not a discussion about 120V or 240V. Form search is back up and upgraded now.

[sloopy]

^^^ You are mixing up numbers, the amperage quoted is at 12v DC (twelve volts direct current) but neither 120v AC nor 240v AC.

Thanks Doh
over-complicate things for myself I did.

Thank you pekatete. You have given me a heaping of great information in your threads.

Dogie, I appreciate you coming in to confirm my mishap. I can still claim n00bness though eh'
Special thanks on search too, that is a huge boon!

[pekatete]

^^^ You are mixing up numbers, the amperage quoted is at 12v DC (twelve volts direct current) but neither 120v AC nor 240v AC.

Thanks Doh
over-complicate things for myself I did.

Thank you pekatete. You have given me a heaping of great information in your threads.

Dogie, I appreciate you coming in to confirm my mishap. I can still claim n00bness though eh'
Special thanks on search too, that is a huge boon!

You are more than welcome .... I try to answer as concisely as I can, whenever I can (without waffling).

[Rabinovitch]

your s3 has a TON of hwe.. it might be worth a repaste.
plus i dont see your difficulty..

update to the 12/19 firmware and see how that works..

if the difficulty is low on the s3 the shares will be low..

Oh, come on!.. According to this formula:

HW / (diffA + diffR + HW) * 100

he has very few HW error rate. 0.00035% if I counted correctly.

[adaseb]

I plugged a lamp into where I have my 2 Antminer S3+ running. Every minute or so the light flickers slightly.

Anyway to solve this? I am assuming this is inrush current with the mining cores changing clock rates or fans adjusting their speeds.

[ybelevi]

hello

i was buy a second hand s3.i did something work with the network settings.(i dont write the broadcast address).thats why icant see login page on gui or cant connect via ssh.butt miner still hashing.i want to reset it but i cant  (maybe for the latest firmware).
what can i do now? how can i reset it? or is there a way to open the login page

thanks for the help

[mrpark]

I hope you are using a 20amp outlet. If its a 15a it really shouldn't be used for mining in my humble opinion.

I plugged a lamp into where I have my 2 Antminer S3+ running. Every minute or so the light flickers slightly.

Anyway to solve this? I am assuming this is inrush current with the mining cores changing clock rates or fans adjusting their speeds.

[marvykkio]

but why bitmain not update the firmware of most antminer s3 +? 09/01/2015 from the many mistakes, now ask you to bitmain but your engineers manage to pull off a firmware good enough?

[aarons6]

but why bitmain not update the firmware of most antminer s3 +? 09/01/2015 from the many mistakes, now ask you to bitmain but your engineers manage to pull off a firmware good enough?

the 12/19 firmware works just fine.

[marvykkio]

yes, now I put the 19/12 and has a lot less mistakes, but with the 09/01/2015 gave more hash,

[CryptoGuy]

I hope you are using a 20amp outlet. If its a 15a it really shouldn't be used for mining in my humble opinion.

I plugged a lamp into where I have my 2 Antminer S3+ running. Every minute or so the light flickers slightly.

Anyway to solve this? I am assuming this is inrush current with the mining cores changing clock rates or fans adjusting their speeds.

15a is fine for 2 miners running. I've got 3 on a single PSU pulling ~9A at the wall. What he needs to do is find out what else is on that breaker. If there's nothing else on the breaker then I would suspect faulty wiring or the breaker itself is going bad.

[aarons6]

yes, now I put the 19/12 and has a lot less mistakes, but with the 09/01/2015 gave more hash,

the amount is insignificant compare to the the newer 1/9 firmware my s3s would show an X sometimes and reboot all by themselves randomly.

[sloopy]

I hope you are using a 20amp outlet. If its a 15a it really shouldn't be used for mining in my humble opinion.

I plugged a lamp into where I have my 2 Antminer S3+ running. Every minute or so the light flickers slightly.

Anyway to solve this? I am assuming this is inrush current with the mining cores changing clock rates or fans adjusting their speeds.

15a is fine for 2 miners running. I've got 3 on a single PSU pulling ~9A at the wall. What he needs to do is find out what else is on that breaker. If there's nothing else on the breaker then I would suspect faulty wiring or the breaker itself is going bad.

I have 3 S3 (not S3+, there is a difference of 340 to 370ish? watts, call it 30 watts per unit) on an EVGA1300 PSU temporarily on a 15 AMP 120vAC line right now. It has been temporarily there for a few days. None of the wire is hot. All connections look good. Good hashrate on the units

I had a hall light doing this same thing, sometimes flickering, it was a crackle, zzttt, zzzZtt, TzTTTZZ and the light would stay on then like nothing. There weren't any miners connected.
I told the kids it was a ghost.

Then I checked the connections at the switch, the fixture, cleaned everything, checked the wire for continuity and it has worked right since then. I've no clue which connection was bad, but it is rock steady now.

The breaker size is important, so is the wire size, and whatever power supply and connectors / wires are between them and your miners. I would always know this before connecting a miner.

If you know what you are doing forgive me for going straight for the basics, but these are the things I always start with when doing anything like this, check the wire and connections, make sure nothing is loose at the wall switch (if there is one), and anywhere it connects. If another device is on the same circuit besides the miner and lamp remove it. Also, try the lamp on a different circuit. I've seen people install the switch upside down and leave it so never assume someone else did it the right way or the way you would if it isn't right

Finally, what size breaker and exactly what devices are on it and if you know what size is the wire? I assume it is US 120vAC?

*I am not an electrician, nor will I pretend to be *)

For more off topic or off color discussion...

I do not do this and do not recommend it. (Hint, use low voltage supply and devices if doing something this stupid anyway) If you do not have a meter you can always do your own redneck continuity testing by using the wire on a different power source and device without anything else on the line! But be warned, do this at your own risk! If you blow something up do not blame me. I do not do this and do not recommend it. You can cause a short if the wire IS cut or broken. It is worth your money to have a meter if you have these kinds of 'toys'! At the same time you can take almost anything I say and make it safe or dangerous in the way you implement. Never ground a live wire, take the extra time, make two trips, know without a doubt the breaker is OFF when it should be OFF.

Invest in a good (I prefer Fluke) meter.  If possible grab a little device called AC Voltage Detector. Slightly bigger than a pen, some bigger or smaller. I picked one up at Lowes and paid too much I'm sure, I think it was 15 USD, but it is made by Southwire, has a not so bright flashlight in it, but what I do like is it is dual colors on the voltage detection end. Red when you are not near power and then green when you are. The red would be so you know the device is on and working along with providing some additional light to see by. The flashlight is on the other end. It doesn't pickup under 50v but goes to 1000V. I'm confident there are better models, but so flippin handy. The red / green end looks like it has a condom on it, cannot miss it.


Sorry for the longwindedness but I got off on that topic.
I do not want to start a new post so not many will read this far.

Anyway about the firmware for the S3 / S3+...

I have been doing my own testing since I have all of the steps down for upgrading / downgrading. It would be great if BITMAIN would work with the people in the community like pekatete and add some of the functionality or at least learn from his testing. For many it isn't always about the best overclock either, it can be about the best running machine at stock.
I have 4 S3+ and 4 S3 I have been running both the 10-24 and 12-19 firmware on. I did start with 1-9 in the mix but did indeed see a drop in hashrate on a unit I upgraded from 12-19. I waited 24 hours, rebooted, and waited another 24. Then I dropped 1-9 from my testing. I wish I hadn't, but since I did it will be a next todo.

The 12-19 firmware is great for the S3 'underperformer' as I call them. The units I have which barely push 440 with normal temps and a 2000w PSU with nothing else on it. Which is the only way I record any 'real' numbers. Changing the voltage in 12-19 does zip for me on the same units where it does have an effect on 10-24. But, the 12-19 seems to manage these units better. Maybe it is something I did in editing frequencies on the 10-24, not sure, but voltage no work for me and I have about the same hash.

10-24 firmware, well it is for everything else. I need an updated 10-24 version to include newer options and cgminer, but adjusting the vcore to the correct setting for the chosen freq does make a big difference and is especially noticeable in the S3+ units and some? S3 with DC2DC... I don't remember but you know what I mean.
The 10-24 firmware gives me overall higher, consistent hashrates  because I am able to make adjustments to the freq and voltage to find the sweetest spot for that particular unit. I keep them clean, fans running, good power, no bs. I want the firmware to be the same

What I would love to see is someone who can look at both 10-24 and 1-9 and give us a hybrid with the good stuff. If anyone needs a unit to test with let me know