ANTMINER S3+ Discussion and Support Thread

[Count_Frackula]

May I ask what determines load?  I have two Antminers,

#1 using cat5 connection and its wifi connection disabled has:
loads of 2.70, 2.56, 2.51,
hashrate 441.03,
DIFF 354,
Fan/Temp 3000/41 & 2220/41 room
Temp/Humidity 83°/-%;

and #2 at some distance running via wifi has:
loads of 1.62, 1.47. 1.42,
hashrate 440.80,
DIFF 367,  
Fan/Temp 2040/40 & 1860/39,
room Temp/Humidity 80°/39%.

So the first has loads 147% higher than the second?  Both S3's at stock frequency.

The biggest load is on an Antminer is cgminer, because it does all the work.
Installing the latest cgminer (which does NOT void your warranty) will, it has been shown, reduce the load by a significant margin.

I am actually surprised at your results, I would have thought the wifi connected antminer would have a higher load than the cat5 connected one, and that it would be reflected in the Diff that has been allocated by the pool, aka I would have expected the cat5 connected ant to have a higher Diff. Could you check your pool-side speeds and see if they reflect the GUI speeds? Also, what are the DiffR numbers for each? (PS. I'll expect a slightly higher DiffR on the wifi connected ant, assuming they are connected to the same pool)

Do you know by chance how much of a load reduction is realized with the new CGMiner?  We've ordered a number of the new S3+ units and am wondering what PSU wattage would be needed for 1 unit pending they've rated it with the older version. I was planning on 500w PSUs for singles... Perhaps 400s would work?

Thanks!

Strato

I wouldn't expect power consumption to be significantly impacted.  The load average is usually on a small router-grade CPU that the units use to run the basic functions.  Power-wise, the core OS probably uses a maximum of about 1 amp (though I suspect it's less than .5 amps).  Less load on this probably affects that power draw by 10% - maybe 20%.  In short, you're talking about a less than a 200 MA difference in power draw, at best.

[1714649566]

1714649566

[1714649566]

1714649566

[1714649566]

1714649566

[1714649566]

1714649566

[1714649566]

1714649566

[pekatete]

Do you know by chance how much of a load reduction is realized with the new CGMiner?  We've ordered a number of the new S3+ units and am wondering what PSU wattage would be needed for 1 unit pending they've rated it with the older version. I was planning on 500w PSUs for singles... Perhaps 400s would work?

Thanks!

Strato

Don't get in a muddle here. Decreasing CPU load does NOT decrease the power consumption of the unit to any significant level.
*** START PINCH OF SALT *** I could argue that it creates "freedom" for the CPU to process even more data thus increase power consumption. *** END PINCH OF SALT ***

I'd say stick to a 500 watt PSU or even better a 550 watt PSU for a single unit.

[soy]

May I ask what determines load?  I have two Antminers,

#1 using cat5 connection and its wifi connection disabled has:
loads of 2.70, 2.56, 2.51,
hashrate 441.03,
DIFF 354,
Fan/Temp 3000/41 & 2220/41 room
Temp/Humidity 83°/-%;

and #2 at some distance running via wifi has:
loads of 1.62, 1.47. 1.42,
hashrate 440.80,
DIFF 367,  
Fan/Temp 2040/40 & 1860/39,
room Temp/Humidity 80°/39%.

So the first has loads 147% higher than the second?  Both S3's at stock frequency.

The biggest load is on an Antminer is cgminer, because it does all the work.
Installing the latest cgminer (which does NOT void your warranty) will, it has been shown, reduce the load by a significant margin.

I am actually surprised at your results, I would have thought the wifi connected antminer would have a higher load than the cat5 connected one, and that it would be reflected in the Diff that has been allocated by the pool, aka I would have expected the cat5 connected ant to have a higher Diff. Could you check your pool-side speeds and see if they reflect the GUI speeds? Also, what are the DiffR numbers for each? (PS. I'll expect a slightly higher DiffR on the wifi connected ant, assuming they are connected to the same pool)

Reading the last few pages I opted to install the latest cgminer on what had been the problem S3, #2.  Before doing so I ran opkg list-installed | grep cgminer on both S3's so as to ascertain the present version.  Both had 3.12.0-1.  I downloaded the latest version on #2 and ran opkg list-installed again.  Right now I'm guessing that installing via wget and rebooting doesn't update the installed listing as after installation of the latest cgminer the installed list still showed 3.12.0-1.  Since that has the lower loads it could be the more efficient miner.

I will install the latest cgminer on #1 and check back a little later today.

The poolside hashrate show #1 to be lower now, GUI 438.38, # 1 having the higher load (2.46) and hardware errors.  #1, cat5 connected, seems to have rebooted 22 minutes ago for no reason, now has a low hardware error count (of course only running 22 minutes) of 5, DIFF 449, DiffA 165394, DiffR 0, DiffS 0.

#2 poolside hashrate higher by 11.7%, GUI 440.28, wifi connected, lower load (1.39), DIFF 367, DiffA 6,986,294, DiffR 5505, DiffS 0.

#2 has new thermal paste applied yesterday.  I'll be picking more up today and might rip #1 apart to change.  Again, I'll download the cgminer to #1 and get back later with new stats.

[pekatete]

Reading the last few pages I opted to install the latest cgminer on what had been the problem S3, #2.  Before doing so I ran opkg list-installed | grep cgminer on both S3's so as to ascertain the present version.  Both had 3.12.0-1.  I downloaded the latest version on #2 and ran opkg list-installed again.  Right now I'm guessing that installing via wget and rebooting doesn't update the installed listing as after installation of the latest cgminer the installed list still showed 3.12.0-1.  Since that has the lower loads it could be the more efficient miner.

I will install the latest cgminer on #1 and check back a little later today.

After updating cgminer, I think there is another way to ascertain the version than the one installed which was mentioned earlier in the thread. If you use the wget method, ensure you chmod+x the new cgminer before rebooting.

The poolside hashrate show #1 to be lower now, GUI 438.38, # 1 having the higher load (2.46) and hardware errors.  #1, cat5 connected, seems to have rebooted 22 minutes ago for no reason, now has a low hardware error count (of course only running 22 minutes) of 5, DIFF 449, DiffA 165394, DiffR 0, DiffS 0.

#2 poolside hashrate higher by 11.7%, GUI 440.28, wifi connected, lower load (1.39), DIFF 367, DiffA 6,986,294, DiffR 5505, DiffS 0.

#2 has new thermal paste applied yesterday.  I'll be picking more up today and might rip #1 apart to change.  Again, I'll download the cgminer to #1 and get back later with new stats.

I'd expect the miner connected via wifi to have a higher DiffR (i.e rejected shares) because of the higher latency inherent in a wifi connection, and your numbers thus far bear this out.
The cat5 connected unit, though showing a lower poolside hash-rate, should have a higher Accepted number (and / or DiffA) compared to the wifi one.
All in all, I would highly recomend updating to the latest cgminer though I have noticed some wierd things since I installed it on my S3's, but in any case, sort your bits and bobs out and get a more accurate picture.

[shadow2]

the s3+ firmware + the 4.6 cg miner has pretty much got rid of the hardware errors I was getting on one of my S3s.  ok it was only 1.5% (according to Mantmonitor) now it's 0.0018%.

you can check the version via the api

if you are on a remote machine, from another copy of cgminer you can do the following command:

Code:

java API version 192.168.1.99

Where 192.168.1.99 is the S3.

Or on the S3 itself run

Code:

cgminer-api version

I have noticed bursts of hash rate significantly over 500 even though my frequencies shouldn't get the rate that high as I'd get an X in the row of os when using the higher frequencies  one is on 237.5 and the other 231.25. just spotted one claiming 657.

[soy]

Reading the last few pages I opted to install the latest cgminer on what had been the problem S3, #2.  Before doing so I ran opkg list-installed | grep cgminer on both S3's so as to ascertain the present version.  Both had 3.12.0-1.  I downloaded the latest version on #2 and ran opkg list-installed again.  Right now I'm guessing that installing via wget and rebooting doesn't update the installed listing as after installation of the latest cgminer the installed list still showed 3.12.0-1.  Since that has the lower loads it could be the more efficient miner.

I will install the latest cgminer on #1 and check back a little later today.

After updating cgminer, I think there is another way to ascertain the version than the one installed which was mentioned earlier in the thread. If you use the wget method, ensure you chmod+x the new cgminer before rebooting.

The poolside hashrate show #1 to be lower now, GUI 438.38, # 1 having the higher load (2.46) and hardware errors.  #1, cat5 connected, seems to have rebooted 22 minutes ago for no reason, now has a low hardware error count (of course only running 22 minutes) of 5, DIFF 449, DiffA 165394, DiffR 0, DiffS 0.

#2 poolside hashrate higher by 11.7%, GUI 440.28, wifi connected, lower load (1.39), DIFF 367, DiffA 6,986,294, DiffR 5505, DiffS 0.

#2 has new thermal paste applied yesterday.  I'll be picking more up today and might rip #1 apart to change.  Again, I'll download the cgminer to #1 and get back later with new stats.

I'd expect the miner connected via wifi to have a higher DiffR (i.e rejected shares) because of the higher latency inherent in a wifi connection, and your numbers thus far bear this out.
The cat5 connected unit, though showing a lower poolside hash-rate, should have a higher Accepted number (and / or DiffA) compared to the wifi one.
All in all, I would highly recomend updating to the latest cgminer though I have noticed some wierd things since I installed it on my S3's, but in any case, sort your bits and bobs out and get a more accurate picture.

Haven't left for shopping yet, busy, but have some stats.

#1 53minutes (58m slow typist)
     436.51GH/s(avg)
     getworks 117
     accepted 911
     rejected 0
     HW 20
     Utility 16
     discarded 4,596
     stale 0
     localwork 703407
     WU 6,122
     DiffA 351659
     BestShare 1,327,130
     load 1.39 1.36 1.29

#2 53m (59min)
     438.76GH/s
     getworks 115
     accepted 1020
     rejected 0
     HW 33
     utility 18
     discarded 5,016
     stale 0
     LocalWork 705,714
     wu 6,143
     DiffA 258,344
     BestShare 205,663
     load 1.32 1.29 1.25

Pool score about the same for each
Pool hashrate a bit lower for #2 but both look good and running well.



    

    






#2

[soy]

the s3+ firmware + the 4.6 cg miner has pretty much got rid of the hardware errors I was getting on one of my S3s.  ok it was only 1.5% (according to Mantmonitor) now it's 0.0018%.

you can check the version via the api

if you are on a remote machine, from another copy of cgminer you can do the following command:

Code:

java API version 192.168.1.99

Where 192.168.1.99 is the S3.

Or on the S3 itself run

Code:

cgminer-api version

I have noticed bursts of hash rate significantly over 500 even though my frequencies shouldn't get the rate that high as I'd get an X in the row of os when using the higher frequencies  one is on 237.5 and the other 231.25. just spotted one claiming 657.
[/quote]

Yes, I see the version number in the download URL.

Wonder what frequency the S3+'s will be shipped as.  Another 11 days or sooner....  And transit will take how long?

[pekatete]

the s3+ firmware + the 4.6 cg miner has pretty much got rid of the hardware errors I was getting on one of my S3s.  ok it was only 1.5% (according to Mantmonitor) now it's 0.0018%.

you can check the version via the api

if you are on a remote machine, from another copy of cgminer you can do the following command:

Code:

java API version 192.168.1.99

Where 192.168.1.99 is the S3.

Or on the S3 itself run

Code:

cgminer-api version

I have noticed bursts of hash rate significantly over 500 even though my frequencies shouldn't get the rate that high as I'd get an X in the row of os when using the higher frequencies  one is on 237.5 and the other 231.25. just spotted one claiming 657.

Hash spikes are not un-usual when your Diff is still low. Let it run for a while and these will become more reasonable (and these will be in the 5s data). Check again after several hours and again after 24 hrs to get a better picture.

PS. Is there any S3 cgminer 4.6 support thread aside from this?

[pekatete]

the s3+ firmware + the 4.6 cg miner has pretty much got rid of the hardware errors I was getting on one of my S3s.  ok it was only 1.5% (according to Mantmonitor) now it's 0.0018%.

you can check the version via the api

if you are on a remote machine, from another copy of cgminer you can do the following command:

Code:

java API version 192.168.1.99

Where 192.168.1.99 is the S3.

Or on the S3 itself run

Code:

cgminer-api version

Yes, I see the version number in the download URL.

Wonder what frequency the S3+'s will be shipped as.  Another 11 days or sooner....  And transit will take how long?

Ahh! you found the post about version checking .... good on you.
As far as the frequesncy of the S3+, check the product web page on bitmaintech, off my head it is 440GH/s but overclockable to 500GH/s.

[soy]

the s3+ firmware + the 4.6 cg miner has pretty much got rid of the hardware errors I was getting on one of my S3s.  ok it was only 1.5% (according to Mantmonitor) now it's 0.0018%.

you can check the version via the api

if you are on a remote machine, from another copy of cgminer you can do the following command:

Code:

java API version 192.168.1.99

Where 192.168.1.99 is the S3.

Or on the S3 itself run

Code:

cgminer-api version

Yes, I see the version number in the download URL.

Wonder what frequency the S3+'s will be shipped as.  Another 11 days or sooner....  And transit will take how long?

Ahh! you found the post about version checking .... good on you.
As far as the frequesncy of the S3+, check the product web page on bitmaintech, off my head it is 440GH/s but overclockable to 500GH/s.

So, the S3+ is the S3 with a firmware upgrade or is there a hardware mod as well or is that information not available?

[pekatete]

So, the S3+ is the S3 with a firmware upgrade or is there a hardware mod as well or is that information not available?

If you'd asked me (I know you did), I'd say the S3+ is the S3 from around batch 6, and ofcourse, you'll also get a firmware upgrade.

[thehairymob]

Tried to use Putty and another SSH program to update CGminer but nope I can't figure it out. Using Putty I connect to the S3 then it won't let me type in password. It is as if my keyboard stops working that is other than the enter key which still works. Am I missing some thing? How do you get the ant to update this vital piece of software?

[PatMan]

Using putty, nothing will appear when you type the password - it's a security thing. Just type the password & hit enter - then hey presto, you're in 

[Pt0x]

I don't have the build environment to create a full binary image, and the code will need a LOT of work to actually become part of mainline cgminer it seems, but for the time being here is a link to a bugfixed binary of cgminer 4.6.0 for the S3 based on the bitmain code. Copy the file into /usr/bin/ on the S3 overwriting the existing one:
http://ck.kolivas.org/apps/cgminer/antminer/s3/4.6.0-140908/cgminer

@CKolivas  Thank you!

I just sent you a small tip to your address.

BTW: I been using the new bin for a couple of hours now without issues.

For those that wan't to update their binary, I used these commands:

 cd /usr/bin/
 mv cgminer cgminer.original
 wget http://ck.kolivas.org/apps/cgminer/antminer/s3/4.6.0-140908/cgminer
mv cgminer cgminer-ants3-4.6.0-140908-kolivas
 chmod +x cgminer-ants3-4.6.0-140908-kolivas
 ln -s cgminer-ants3-4.6.0-140908-kolivas cgminer
/etc/init.d/cgminer restart

This way, the reboot is not necessary + you can keep a good name to describe the new binary.

Cheers!

[gonzo_ucab]

Ours Bitmaint setup. 7Ths There.

http://pic.twitter.com/XOMtJvHQOH

Happy mining everyone

[not.you]

Anyone know if that most recent firmware that says it is compatible with the S3+ also works on the earlier batches?  I haven't kept up with this thread the whole time but I have a batch 2 that has never done very well and I wondered if that might help it out.

[ccollectables]

I have bricked a S3 with an improperly downloaded frimware upgrade,  does anyone know if there is a way to reflash the firmware or to do a hard reset (holding reset button did not work) without being able to access the GUI?

[Hunterbunter]

Has anyone received their refund on the B1-3's yet? It's been months and nothing. The website just says status "OK"...what is that supposed to mean?

[grn]

Has anyone received their refund on the B1-3's yet? It's been months and nothing. The website just says status "OK"...what is that supposed to mean?

loooong ago and applied it to B3

[suchmoon]

Has anyone received their refund on the B1-3's yet? It's been months and nothing. The website just says status "OK"...what is that supposed to mean?

Have you checked your wallet? Refunds were issued in July.