Topic: Antminer S11 19 TH not reaching 19 th (Read 239 times)

No, you clearly do not understand that the numbers shown can be high or low over a short period of time.
5 mins in CERTAINLY within the realms of a short period of time.

Heck I've seen a miner show 150% for an hour due to luck finding low diff shares.

You can even work out the probabilities of that - since it's EXACTLY the same working as the probabilities with block finding.
Miners haven't mined at 1diff since before the S2, so the difficulty produced shown on the web page is based on a much larger value also.

---

Edit: from the POV of a pool: if a miner mines at 19THs, what's the chance that for 90 shares, it will average worse than 18.2THs?
This is calculated using the probability density function of the Erlang distribution.
On my pool web site I call this the CDF[Erl] and show that for various pool block statistics.

My pool sets 18 shares per minute, so 90 shares would normally take 5 minutes at 19THs or 3.333 seconds per share.
If a miner, with the same 19THs specs, instead was averaging 18.2THs, it would take 19/18.2 times as long.
i.e. 104.4% or about 5 1/4 minutes
So what's the chance of 90 shares averaging 104.4% - CDR[Erl] of 90, 1.044 = 0.672371
So that means a 1 in (1 / (1-0.672371)) chance of being worse than 18.2THs = 1 in 3.05 chance - yeah pretty often.

From the POV of the miner it depends on the difficulty it mines internally, you'd have to find that out.
But lets say it does something like 1024 diff
At 1024 Diff a 19THs miner would be expected to find 19*10^12 / (1024 * 2^32) = 4.32 1024 Diff shares per second
So for 5 minutes that's ~1296 shares.
So what's the CDF[Erl] of 1296, 1.044? 0.941859
So that means a 1 in (1 / (1-0.941859)) chance of being worse than 18.2THs = 1 in 17.2 chance

i.e. you run 17 miners, or start your miner 17 times, you'd expected to get one to close to 18.2THs or worse for the first 5 minutes.

The pool Diff for a 19THs miner is 19*10^12 * (60/18) / (2^32) = 14746 Diff
Bitmain miners mine at a power of 2 - so the next power of 2 below that would be 8192 so if that is what it is mining at, then it's more likely than 1 in 17.2 by a lot.

His picture shows the pool Diff higher (65536) so either he set it himself or the pool doesn't do it too well.

That is rather a stupid comparison to say the least, we are not discussing how much BTC will that s11 mine over the next 12 months, we are discussing about the current figures.

Your theory about a chain's hashare increasing overtime is like saying if you let your faulty 1600cc engine idle for a few hours it turns to a brand new 2000cc engine. (that's exactly how stupid you it sounds)

I know you have nothing solid to support your false claims, so you will keep wasting my time, but i rather go do something else. nothing personal tho.

regards.

Sigh - as I said - above your pay grade.

Go ask one person at random who will be the next President of your country.

That of course means that is who will be the next president? lol

How many do you have to ask? What is the confidence interval of how many you do ask?

Guess what, even the media people who do the polls don't know the answer to those important questions

11 seconds is not 5 mins, it was only there to show you that the freq is the same and ideal hashrate per chain is set at second 0 , so if it says 5,6th  it won't go to 6.8th, not even in a millennium.

you just had a newbie wait for hours for something that is not going to change , well done  


you have not explained anything, you pulled a false piece of info outta your ass and i corrected it with a proof and of course you can't live with the fact that someone has pointed out your b.s.

TL;DR : 5 mins is more than enough time to get a clear idea on what hashrate should you expect from the miner, now unless you have a valid proof that proves the opposite ( some screenshots the show a board starting with freq of 600 and runs for 5 mins then magically move up to 725)  then stop spreading wrong information. but i know you won't  

Hmm your 11 seconds was low ... I guess that's not possible?

It's a statistical issue - as I explained already - but obviously above your pay grade.

P.S. I've installed a few thousand S9s for various people ...

i had to reboot one of my gears to prove this

5 days :



11 seconds:



5 mins :



8 mins :



notice how the frequency is set during start up, as well is the ideal hashrate ,it took less than 5 mins to reach the to almost the exact same RT hashrate, it took about 8 mins or a bit more to get to the same average and that's normal giving the fact that the miner hashrate will start at almost 0 and then increase.

Give us 1 reason why would that particular board start will a low frequency when others do not? or why would it increase after running for a few hours or days?

Measuring hashrate on the pool side needs a longer period of time ( i am not disagreeing to this), however the hashrate reported by the miner almost instantly becomes stable.

I speak from personal experience , i watched tons of miners start up and usually whatever you see in the first a few minutes is all you going to get.

if it is a firmware issue, then a simple reboot will change the results, aside from that, the frequency won't increase neither the hashrare.
 
i am certain that if he does not reboot the miner, that hashrate won't increase much  , not even in a year. but let's see what he posts next.

He does not actually mention about SSH but using 2019 firmware of any antminer machines will give you result "cannot find signature" and disabled SSH access. This is a common issue of latest firmware and you can't revert it back to a lower firmware version unless if he can disable the SSH he can bypass the "cannot find signature" issue.

His final solution here is to flash the s11 control board with SD card program recovery but I don't know if the released SD card firmware from bitmain is the lower version of s11.

Wrong.

It's a 5 minute sample of mining - 5 minutes means next to nothing statistically since the sample is about 90 shares.

Shares are random with exactly the same random properties as finding bitcoin blocks.

Once he comes back with having run it for a couple of hours, then he might have something worth checking.


Nope, you're not reading the thread.
He said his previous SSH-able firmware said under 19TH also ... ... ... ... ...

First of all, change to a different pool , even for testing.

with all due respect to the members who "think" more time "might" increase the hashrate , i am almost certain that's pretty much all you going to get, and it could only get worse.

It's pretty clear that your first chain is not doing so great, 2 hw errors in 5 mins , frequency is stuck at only 600, i suspect the pool results will look a bit lower with all those hw errors you going to get from the first chain.

it could be a firmware issue, but it's most likely a hardware issue.

can you try to switch the data cable as well as the power cable from 1st to 2nd board (not sure if you can do this without disassembling something on the s11 so DYOR )

That would be the only way to know if it's a power/data cable related issue or the it's just a bad board.

** a different firmware might slightly increase the hashrate on that board, but it's very likely not going to hash similar to the other two chains.

For what this thread is about, you are completely missing the point. Bitmain has removed the ability to downgrade the firmware as well as SSH access. If you still fail to see the consequences of that, feel free to reply in the other thread, but then again you already did so i don't see your point in further arguing this.

Fact is, he as the S11 owner should have had the choice to downgrade, as it has always been until the 2019 firmware. This was like the PS3 fiasco when Sony later removed the "Other OS" feature that allowed running a regular Linux. Since the 2019 firmware is taking away features already present when bought originally.

19TH miner mining at 18.2TH

Elapsed 5m32s

Seriously? You gonna blame lack of SSH, when the reason is using a poor sample to decide the hash rate of a miner?

Seriously your reply is complete rubbish and no idea why you'd even say it ...

And yet another reason to AVOID Bitmain's firmware from 2019. How is this move not considered hostile? It is a serious blow to Bitmain's customer base. So oops, we made a mistake, but now you can't ever revert back to the older working firmware?

I hope OP finds a way, as convoluted as it may be, to revert back to the older firmware. Something that should have been a simple "lets check if the older version works" has now become impossible thanks to a new artificial barrier erected by Bitmain.

I agree philipma... the picture show problem on only 1 hashboard ... the problem comw with the defect psu , or just the hashboard is bad perf... the firmware adjust the freq with better efficiency but if the hashboard is bad or down, the firmware lower this freq for more stability

yeah  you should have asked us first before loading firmware.

5 min tests are a waste of time.

2, 3, 4 even 12 hour tests are better.

So  next time try 3 long tests on 3 different pools  then ask.

As for now  my thoughts are you have an underperforming unit but since you did not do proper tests before changing the firmware it is hard to tell.

ant pool may have had an issue
the cable you used for that unit may have a problem
the port on your switch could be so so
the controller on that miner may be so so
the psu on that miner may be so so
the hash boards on that miner may be so so

I see on low board on that unit most likely you have a poor board but once you alter firmware all bets are off.

how about your other units did you alter the firmware on them to march of 2019?

Hi there,

One of my S11 units is not reaching 19TH/s, it is hovering around 18.2.

I tried resetting many times and also upgrading to latest firmware, it was actually hovering around 18.6 before I upgraded lol.

It is currently running: Antminer-S11-all-user-201903191429-sig.tar.gz

First I tried to load: Antminer-S11-xilinx-201812082324-autofreq-user-Update2UBI-NF.tar.gz - however it said cannot find signature.

Any thoughts?

Thanks in advance.

https://imgur.com/a/Y9AQTwi