ANTMINER S3+ Discussion and Support Thread

[OrangeGator]

Yeah, I know.  I tried them with 1 miner per EVGA psu, plus the third on an old Corsair TX750.  Still no luck overclocking.  I am using a Kill a Watt, and am measuring 1080W at the wall at stock speeds.  The max it ever hit was 1150W when trying to overclock.

Well how are you going to overclock 3 S3's with 2 power supplies that can't supply enough current to do it??
Each S3 at 250 will draw 400+ watts.
I would connect just 1 or 2 and try to overclock them 243 and 250. if they work then it is what I noted, the power supplies cant supply the amps needed for 3 overclocked

[1714624679]

1714624679

[1714624679]

1714624679

[1714624679]

1714624679

[MissouriMiner]

Yeah, I know.  I tried them with 1 miner per EVGA psu, plus the third on an old Corsair TX750.  Still no luck overclocking.  I am using a Kill a Watt, and am measuring 1080W at the wall at stock speeds.  The max it ever hit was 1150W when trying to overclock.

Well how are you going to overclock 3 S3's with 2 power supplies that can't supply enough current to do it??
Each S3 at 250 will draw 400+ watts.
I would connect just 1 or 2 and try to overclock them 243 and 250. if they work then it is what I noted, the power supplies cant supply the amps needed for 3 overclocked

Here are some stats I gathered, just for information sharing.  As many are aware, the power from the wall depends on PSU efficiency and used capacity (% of MAX).  A single 1000W PSU powering 2x S3+'s will draw more power from the wall than 2 separate same model 1000W PSU's. The higher the PSU capacity , the lower the efficiency.

MHz Setting   Avg Gh/s   Watts      "Efficiency W/Gh"      PSU Model         

225            453           400           0.88                  1000W Coolmax, 80 Plus Bronze         

225            453           365           0.81                  1050W Cougar GX1050, 80 Plus Gold         

225            453           390           0.86                  1200W Coolmax, 80 Plus Gold
237            470           415           0.88                  (Running 2x S3+, Values are per S3+)               
250            500           445           0.89            

225            453           365           0.81                  1300W Rosewill Lightening, 80 Plus Gold
237            475           385           0.81                  (Running 2x S3+, Values are per S3+)      
250            500           405           0.81            

[yslyung]

Yeah, I know.  I tried them with 1 miner per EVGA psu, plus the third on an old Corsair TX750.  Still no luck overclocking.  I am using a Kill a Watt, and am measuring 1080W at the wall at stock speeds.  The max it ever hit was 1150W when trying to overclock.

Well how are you going to overclock 3 S3's with 2 power supplies that can't supply enough current to do it??
Each S3 at 250 will draw 400+ watts.
I would connect just 1 or 2 and try to overclock them 243 and 250. if they work then it is what I noted, the power supplies cant supply the amps needed for 3 overclocked

Here are some stats I gathered, just for information sharing.  As many are aware, the power from the wall depends on PSU efficiency and used capacity (% of MAX).  A single 1000W PSU powering 2x S3+'s will draw more power from the wall than 2 separate same model 1000W PSU's. The higher the PSU capacity , the lower the efficiency.

MHz Setting   Avg Gh/s   Watts      "Efficiency W/Gh"      PSU Model         

225            453           400           0.88                  1000W Coolmax, 80 Plus Bronze         

225            453           365           0.81                  1050W Cougar GX1050, 80 Plus Gold         

225            453           390           0.86                  1200W Coolmax, 80 Plus Gold
237            470           415           0.88                  (Running 2x S3+, Values are per S3+)               
250            500           445           0.89            

225            453           365           0.81                  1300W Rosewill Lightening, 80 Plus Gold
237            475           385           0.81                  (Running 2x S3+, Values are per S3+)      
250            500           405           0.81            

thx for sharing, might be a noob question ... if power is 230V compared to 110V would it make a difference ? i think it would.

[MissouriMiner]

Yeah, I know.  I tried them with 1 miner per EVGA psu, plus the third on an old Corsair TX750.  Still no luck overclocking.  I am using a Kill a Watt, and am measuring 1080W at the wall at stock speeds.  The max it ever hit was 1150W when trying to overclock.

Well how are you going to overclock 3 S3's with 2 power supplies that can't supply enough current to do it??
Each S3 at 250 will draw 400+ watts.
I would connect just 1 or 2 and try to overclock them 243 and 250. if they work then it is what I noted, the power supplies cant supply the amps needed for 3 overclocked

Here are some stats I gathered, just for information sharing.  As many are aware, the power from the wall depends on PSU efficiency and used capacity (% of MAX).  A single 1000W PSU powering 2x S3+'s will draw more power from the wall than 2 separate same model 1000W PSU's. The higher the PSU capacity , the lower the efficiency.

MHz Setting   Avg Gh/s   Watts      "Efficiency W/Gh"      PSU Model         

225            453           400           0.88                  1000W Coolmax, 80 Plus Bronze         

225            453           365           0.81                  1050W Cougar GX1050, 80 Plus Gold         

225            453           390           0.86                  1200W Coolmax, 80 Plus Gold
237            470           415           0.88                  (Running 2x S3+, Values are per S3+)               
250            500           445           0.89            

225            453           365           0.81                  1300W Rosewill Lightening, 80 Plus Gold
237            475           385           0.81                  (Running 2x S3+, Values are per S3+)      
250            500           405           0.81            

thx for sharing, might be a noob question ... if power is 230V compared to 110V would it make a difference ? i think it would.

It doesn't matter.  This is regarding Watts, which is constant between 110V and 220V.   With 220V, you are using double the Volts at half the Amps.  But the Watts is the same.  Volts * Amps = Watts

[edonkey]

After little less than 80 days, my B3 S3s have broken even. They've mined as much BTC as they cost (factoring in the refund). This is with free power.

There really is no deal like this out there now. Everything is too expensive.

Bitmain, I'm sure that you're not moving units like you were before. I understand that you need to make money. And clearly BTC's price isn't helping.

But if you want to move units, you have to cut prices. If the S4 had a chance of a 90 day ROI, I'd buy today. And I'm sure I'm not the only one sitting on the sidelines waiting for a reasonable deal to come around.

[Rabinovitch]

It doesn't matter.  This is regarding Watts, which is constant between 110V and 220V.   With 220V, you are using double the Volts at half the Amps.  But the Watts is the same.  Volts * Amps = Watts

A little offtopic, but every child in exUSSR knows it from the school curriculum... 

[Dallski]

Yeah, I know.  I tried them with 1 miner per EVGA psu, plus the third on an old Corsair TX750.  Still no luck overclocking.  I am using a Kill a Watt, and am measuring 1080W at the wall at stock speeds.  The max it ever hit was 1150W when trying to overclock.

Well how are you going to overclock 3 S3's with 2 power supplies that can't supply enough current to do it??
Each S3 at 250 will draw 400+ watts.
I would connect just 1 or 2 and try to overclock them 243 and 250. if they work then it is what I noted, the power supplies cant supply the amps needed for 3 overclocked

Here are some stats I gathered, just for information sharing.  As many are aware, the power from the wall depends on PSU efficiency and used capacity (% of MAX).  A single 1000W PSU powering 2x S3+'s will draw more power from the wall than 2 separate same model 1000W PSU's. The higher the PSU capacity , the lower the efficiency.

MHz Setting   Avg Gh/s   Watts      "Efficiency W/Gh"      PSU Model         

225            453           400           0.88                  1000W Coolmax, 80 Plus Bronze         

225            453           365           0.81                  1050W Cougar GX1050, 80 Plus Gold         

225            453           390           0.86                  1200W Coolmax, 80 Plus Gold
237            470           415           0.88                  (Running 2x S3+, Values are per S3+)               
250            500           445           0.89            

225            453           365           0.81                  1300W Rosewill Lightening, 80 Plus Gold
237            475           385           0.81                  (Running 2x S3+, Values are per S3+)      
250            500           405           0.81            

thx for sharing, might be a noob question ... if power is 230V compared to 110V would it make a difference ? i think it would.

It doesn't matter.  This is regarding Watts, which is constant between 110V and 220V.   With 220V, you are using double the Volts at half the Amps.  But the Watts is the same.  Volts * Amps = Watts

Actually, it has been shown that power supplies using 208-240V rather than 110/120V are 1% more efficient. Aside from 110/220, there are two other important efficiency factors: (1) Ambient Temperature and (2) Percentage of PSU capacity used. With the first, I have seen a 1.3% difference in power consumption between 60*F and 80*F ambient temperatures, mostly due to the fans having to work more or less. With the percentage of PSU capacity used, the sweet spot of most PSU's is 40%. That is where they are most efficient. Efficiency curves are not flat. And Power Supply ratings are based on output, not input, so you can have a 500W bronze PSU pull 600W at the wall. Try actually running one S3 on that Rosewill Lightning and see if you can still hit those same numbers. Then try running three. I bet you will have different results.

[MissouriMiner]

Yeah, I know.  I tried them with 1 miner per EVGA psu, plus the third on an old Corsair TX750.  Still no luck overclocking.  I am using a Kill a Watt, and am measuring 1080W at the wall at stock speeds.  The max it ever hit was 1150W when trying to overclock.

Well how are you going to overclock 3 S3's with 2 power supplies that can't supply enough current to do it??
Each S3 at 250 will draw 400+ watts.
I would connect just 1 or 2 and try to overclock them 243 and 250. if they work then it is what I noted, the power supplies cant supply the amps needed for 3 overclocked

Here are some stats I gathered, just for information sharing.  As many are aware, the power from the wall depends on PSU efficiency and used capacity (% of MAX).  A single 1000W PSU powering 2x S3+'s will draw more power from the wall than 2 separate same model 1000W PSU's. The higher the PSU capacity , the lower the efficiency.

MHz Setting   Avg Gh/s   Watts      "Efficiency W/Gh"      PSU Model         

225            453           400           0.88                  1000W Coolmax, 80 Plus Bronze         

225            453           365           0.81                  1050W Cougar GX1050, 80 Plus Gold         

225            453           390           0.86                  1200W Coolmax, 80 Plus Gold
237            470           415           0.88                  (Running 2x S3+, Values are per S3+)               
250            500           445           0.89            

225            453           365           0.81                  1300W Rosewill Lightening, 80 Plus Gold
237            475           385           0.81                  (Running 2x S3+, Values are per S3+)      
250            500           405           0.81            

thx for sharing, might be a noob question ... if power is 230V compared to 110V would it make a difference ? i think it would.

It doesn't matter.  This is regarding Watts, which is constant between 110V and 220V.   With 220V, you are using double the Volts at half the Amps.  But the Watts is the same.  Volts * Amps = Watts

Actually, it has been shown that power supplies using 208-240V rather than 110/120V are 1% more efficient. Aside from 110/220, there are two other important efficiency factors: (1) Ambient Temperature and (2) Percentage of PSU capacity used. With the first, I have seen a 1.3% difference in power consumption between 60*F and 80*F ambient temperatures, mostly due to the fans having to work more or less. With the percentage of PSU capacity used, the sweet spot of most PSU's is 40%. That is where they are most efficient. Efficiency curves are not flat. And Power Supply ratings are based on output, not input, so you can have a 500W bronze PSU pull 600W at the wall. Try actually running one S3 on that Rosewill Lightning and see if you can still hit those same numbers. Then try running three. I bet you will have different results.

Yes, I agree with the PSU capacity impact, which is why I said the same thing.  I agree with the temperature impact too, but am not testing that aspect.  I've heard disagreements, and back and forth points regarding efficiency of 220V vs. 110V.  Some people (not me) say it is more related to the quality of the wiring.  I can't speak to the voltage point and don't disagree or agree.  I'm working on 1x and 3x S3 numbers on the Rosewill, and will share the results.  One problem is that I don't have 2 S3's that will do 250MHz.  Most of mine top out at 237 MHz.  So I'll be short some metrics.  I also have an Antec 1300W Platinum, but it is currently in an S2 and I'm not sure if I want to mess with it.

[Mudbankkeith]

I like the new C1 concept.
It's just a shame it is only approx 1Gh at 800watts.

If it was 2Gh at 800/1000 watts it would maybe return a ROI.

I could see a bank of these connected into my central heating system, all they need is a pump and an indirect coil in the water cylinder.

[IITravel01]

Yeah, I know.  I tried them with 1 miner per EVGA psu, plus the third on an old Corsair TX750.  Still no luck overclocking.  I am using a Kill a Watt, and am measuring 1080W at the wall at stock speeds.  The max it ever hit was 1150W when trying to overclock.

Well how are you going to overclock 3 S3's with 2 power supplies that can't supply enough current to do it??
Each S3 at 250 will draw 400+ watts.
I would connect just 1 or 2 and try to overclock them 243 and 250. if they work then it is what I noted, the power supplies cant supply the amps needed for 3 overclocked

Here are some stats I gathered, just for information sharing.  As many are aware, the power from the wall depends on PSU efficiency and used capacity (% of MAX).  A single 1000W PSU powering 2x S3+'s will draw more power from the wall than 2 separate same model 1000W PSU's. The higher the PSU capacity , the lower the efficiency.

MHz Setting   Avg Gh/s   Watts      "Efficiency W/Gh"      PSU Model         

225            453           400           0.88                  1000W Coolmax, 80 Plus Bronze         

225            453           365           0.81                  1050W Cougar GX1050, 80 Plus Gold         

225            453           390           0.86                  1200W Coolmax, 80 Plus Gold
237            470           415           0.88                  (Running 2x S3+, Values are per S3+)               
250            500           445           0.89            

225            453           365           0.81                  1300W Rosewill Lightening, 80 Plus Gold
237            475           385           0.81                  (Running 2x S3+, Values are per S3+)      
250            500           405           0.81            

thx for sharing, might be a noob question ... if power is 230V compared to 110V would it make a difference ? i think it would.

It doesn't matter.  This is regarding Watts, which is constant between 110V and 220V.   With 220V, you are using double the Volts at half the Amps.  But the Watts is the same.  Volts * Amps = Watts

Actually, it has been shown that power supplies using 208-240V rather than 110/120V are 1% more efficient. Aside from 110/220, there are two other important efficiency factors: (1) Ambient Temperature and (2) Percentage of PSU capacity used. With the first, I have seen a 1.3% difference in power consumption between 60*F and 80*F ambient temperatures, mostly due to the fans having to work more or less. With the percentage of PSU capacity used, the sweet spot of most PSU's is 40%. That is where they are most efficient. Efficiency curves are not flat. And Power Supply ratings are based on output, not input, so you can have a 500W bronze PSU pull 600W at the wall. Try actually running one S3 on that Rosewill Lightning and see if you can still hit those same numbers. Then try running three. I bet you will have different results.

Most power supplies will be about 2% more efficient at 230v vs. 115v (actually a range of 110v to 120v).  My Gold power supply is about 90.8% efficient at 120v and 92.7% at 230v according to the power supply display (running 2 S3+'s at 237.5M = about 778watts on the ThorTech Thunderbolt Plus 1200w power supply display).  Anyone done a graph on what the peak efficiency of the S3 is (from 100M-250M)?  I've got all the equipment here to do the tests that I was thinking of doing this weekend.  I've got 2 S3+'s hooked up the a Corsair AX1500i Titanium power supply (the most efficient power supply you can buy) and with 2 it's at about its peak efficiency point of 50-60% load.  I've got a 3000watt step of voltage converter/transformer (problem in the U.S. is even if the 230v is more efficient, you have to convert it here and that takes up some loss in efficiency vs. those where it's 230v at the wall.  My 3,000watt step up converter takes about 40watts at idle and goes down to 20watts when using with the 2 S3+'s at 237.5M.  So the power for the system went from 820watts at the wall to 840watts with the 230v step up converter was placed in the line).  I've also got a 120v kill-a-watt (for power at the wall) and a 230v kill-a-watt (arrived today, for power after the 230v step up converter/transformer) as well as a 1,000watt/1,500VA sine wave UPS (230v version).  Should give a pretty accurate reading for an efficiency graph if someone wants to put one together.

[Ilan123]

I like the new C1 concept.
It's just a shame it is only approx 1Gh at 800watts.

If it was 2Gh at 800/1000 watts it would maybe return a ROI.

I could see a bank of these connected into my central heating system, all they need is a pump and an indirect coil in the water cylinder.

I would still prefer to buy 2 x S3's after all the additional hassle with purchasing the cooling Kit and extra costs involved etc. Apparently GAW Miners LLC have commissioned 5 Petahash of C1's to be manufactured for them.

[Mudbankkeith]

I like the new C1 concept.
It's just a shame it is only approx 1Gh at 800watts.

If it was 2Gh at 800/1000 watts it would maybe return a ROI.

I could see a bank of these connected into my central heating system, all they need is a pump and an indirect coil in the water cylinder.

I would still prefer to buy 2 x S3's after all the additional hassle with purchasing the cooling Kit and extra costs involved etc. Apparently GAW Miners LLC have commissioned 5 Petahash of C1's to be manufactured for them.

0.8w/Gh is old tech now.

unless there is totally free electric they will soon be out of the game.

[Gogreen]

I prefer 0.5 or 0.48w/gh  to combat the electric bill 

[Biodom]

I prefer 0.5 or 0.48w/gh  to combat the electric bill  

In my area price on electricity went down a bit, which is nice.
Just 6c/kwh plus ~4c/delivery/kwh comparing with current 6.7c/kwh and same for delivery.
I read somewhere that electricity prices are projected to decline 24-36 mo out.

[Mudbankkeith]

maybe there will soon be a C2 at 2Gh and 800watts?

[klondike_bar]

maybe there will soon be a C2 at 2Gh and 800watts?

0.4w/GH? unlikely.

More likely is an S2-style miner that packs 4TH/3kW into a 4U case, with 4 separate coolant loops.

[Xian01]

Anecdote: I noticed overnight that all my S3's failed to honor the failover settings when the primary pool went "Dead". (Running latest-and-greatest Bitmain S3+ firmware)

Woke up to all my S3's sitting cool and idle this morning. S2's were hashing away like champs running Kano's K2 firmware, honoring the failover settings.

Hitting "Save and Apply" on the miner configuration page got things working well again on the S3's switching to the next available server.

[-ck]

Here's an updated S3 binary.

http://ck.kolivas.org/apps/cgminer/antminer/s3/4.6.1-141020/cgminer

Recommended if you're mining on p2pool for the default binary actually discards stale shares which you should never do, especially on p2pool. Also includes changes to queuing and memory usage that were necessary on S4 but probably only of minor benefit here. Recommend you edit the cgminer startup script to remove the --queue value entirely, and add --lowmem. Performance should be pretty much unchanged.

[IYFTech]

Here's an updated S3 binary.

http://ck.kolivas.org/apps/cgminer/antminer/s3/4.6.1-141020/cgminer

Recommended if you're mining on p2pool for the default binary actually discards stale shares which you should never do, especially on p2pool. Also includes changes to queuing and memory usage that were necessary on S4 but probably only of minor benefit here. Recommend you edit the cgminer startup script to remove the --queue value entirely, and add --lowmem. Performance should be pretty much unchanged.

Excellent ck! Will try it now, many thanks 

[jelin1984]

Any updated for s2 miner my friend