Announcement: Bitmain launches AntMiner solution, 0.68 J/GH on chip

[klondike_bar]

Who is Bitmain and what is AntMiner?

I haven't really been following Bitmain.  Maybe someone can explain how it is that they can spend a fortune to develop their own chips and then only sell a few units at a time.  Certainly they should be flooding the market with their devices, no?

I don't know what it is but even though they are actually selling real devices, they still seem scammy to me.  Are these really their own chips or just Bitfurys in disguise?

sushi would be able to say more; but heres my take.

Bitmain is a chinese group that presumably opted to create a working ASIC device before making it public, so that they could show themselves as a team capable of delivery. I got the Antminer from the first auction, and it shipped very quick using an expedited UPS service. Build quality is better than i expected (see the reviews on the past few pages) and is possibly the best delivered product so far in terms of build quality ive seen (i have not actually seen a BFL firsthand, but those probably are the overall winner for cleanest looking device). The funding needed to produce just 30 or 100 units at a time is likely very high, and the first few batches are in effect the beta testing. I would expect some simple but effective changes to be made in later batches

The chips are thier own. In terms of power-efficiency, they are middle-of-the-road, a little worse than bitfury or avalon chips running at the same speed.

[sushi]

Some of the photos to include the PCBs and how it look are here: https://bitcointalk.org/index.php?topic=344970.0

I don't think it's a bootleg product.  Power hungry came from running the chip at 2.8GH and their specs shows less power consumption if the hashing rate was lower.

They are more than capable of running 30-50 units per day.  I was at their production lines and when the they were pushing these out in full speed,  I could feel the ground shaking and it was actually fun looking at these finished boards coming out of the automation lines one after another. It's not a hand solder PCBs or chips.  These were done in the Japanese Brand equipment in the automation process with line of people doing the quality control by both visual and tester checks on each PCBs.  I think I can say is they are believing in the quality control to stay in for the long term.

It does hash away as they claim on the product spec.  I have one here and it does work.  If it came with the PSU it would even be better, but they could release these with PSU with a few weeks of delay.  ( I think the miners want the unit if it's hashing rather than waiting for a plug & play PSU)  Sooner we can have the miner is better of we are.

I felt like I was a kid in the candy store.

They have documentations in gethub i believe.  I know others speculated it was the re-branded avalon but i think it was proved false.

[xstr8guy]

Who is Bitmain and what is AntMiner?

I haven't really been following Bitmain.  Maybe someone can explain how it is that they can spend a fortune to develop their own chips and then only sell a few units at a time.  Certainly they should be flooding the market with their devices, no?

I don't know what it is but even though they are actually selling real devices, they still seem scammy to me.  Are these really their own chips or just Bitfurys in disguise?

sushi would be able to say more; but heres my take.

Bitmain is a chinese group that presumably opted to create a working ASIC device before making it public, so that they could show themselves as a team capable of delivery. I got the Antminer from the first auction, and it shipped very quick using an expedited UPS service. Build quality is better than i expected (see the reviews on the past few pages) and is possibly the best delivered product so far in terms of build quality ive seen (i have not actually seen a BFL firsthand, but those probably are the overall winner for cleanest looking device). The funding needed to produce just 30 or 100 units at a time is likely very high, and the first few batches are in effect the beta testing. I would expect some simple but effective changes to be made in later batches

The chips are thier own. In terms of power-efficiency, they are middle-of-the-road, a little worse than bitfury or avalon chips running at the same speed.

Thanks for bringing me up to speed!

[-ck]

https://github.com/bitmaintech/cgminer

That appears to be bitmains fork of cgminer. I think you dont need full firmware source to modify cgminer... just simple cross compile and replacing the binary should be sufficient.... Just guessing no past experience with openwrt/mips

Having played with this code to try and bring it into mainline cgminer, now it appears this is old non-working code and does not match the code distributed with the device. I have pinged them to try and get the most up to date code as I can't get this one working.

[sushi]

Bitmain AntMiner S1 180GH/S Group Buy for 50 Units

BTC 4.25 each to include Free World-Wide Shipping

Ship on December 3, 2013

14 Units In-Stock Remaining...

https://bitcointalk.org/index.php?topic=353650.0

[ktzhan]

I would at least appreciate some optimizations to the AntMiner port of cgminer, see below statistics of 2 Bitmain AntMiners running in a pool for 36 hours+:

https://i.imgur.com/5if7pET.jpg

Aexcu, it seems that there is something wrong with your miner. the number of HW is more than the number of Accepted?! it should be abnormal. For antminer, the number of HW is near zero. Please check your power supply at first.

[ktzhan]

One of my units perform lower than expected.

https://i.imgur.com/458LEFMl.png

Onboarded my last unit today, and it underperformes significantly.

It was doing ~140 GH/s . Chain 1 having some x , chain 2 many x after several hours of running. I forgot to take a screenshot then. Thinking this could be a PSU issue, i swapped PSU from a fully functional unit, and still see some x. not as low as earlier, but im sure after couple of hours it will get the many x in chain 2... Will update with fresh screenshot if i spot it again...

My ambient : 20 to 30C (may get higher during mid-day and if AC is off).

Any ideas? Anything I can do to see some debug logs? Is cgminer logs saved anywhere? Im very comfortable with linux, ssh and command line...

Is this unit broked?

turtle83 , for the first batch, the there is some bugs for ASIC status checking. If you doubt the ASIC status, please contact our sales/supoort to update your control board.

[-ck]

I would at least appreciate some optimizations to the AntMiner port of cgminer, see below statistics of 2 Bitmain AntMiners running in a pool for 36 hours+:

Aexcu, it seems that there is something wrong with your miner. the number of HW is more than the number of Accepted?! it should be abnormal. For antminer, the number of HW is near zero. Please check your power supply at first.

Wrong. 5.5 million diff1 equivalent shares versus 13.5 thousand hardware errors. You're making a classic mistake of comparing the wrong values. His hardware error count is <.3%

[bclcjunkie]

ckolivas,

thanks for clarifying that... i was worried my 2 miners HW count was getting high...

based on my HW count, one of them is exceeding 0.3 is that something i can ignore?

810/781798 * 100% = 0.10% which is still lower than 0.3 right
3225/773226 * 100% = 0.41 which is higher than 0.3

I would at least appreciate some optimizations to the AntMiner port of cgminer, see below statistics of 2 Bitmain AntMiners running in a pool for 36 hours+:

Aexcu, it seems that there is something wrong with your miner. the number of HW is more than the number of Accepted?! it should be abnormal. For antminer, the number of HW is near zero. Please check your power supply at first.

Wrong. 5.5 million diff1 equivalent shares versus 13.5 thousand hardware errors. You're making a classic mistake of comparing the wrong values. His hardware error count is <.3%

[nexus99]

I want one of these!

[-ck]

ckolivas,

thanks for clarifying that... i was worried my 2 miners HW count was getting high...

based on my HW count, one of them is exceeding 0.3 is that something i can ignore?

810/781798 * 100% = 0.10% which is still lower than 0.3 right
3225/773226 * 100% = 0.41 which is higher than 0.3

I don't know what the normal expected range is for this hardware. Most ASICs, anything under 2% is good, but if hardware errors are expected to be already filtered out by the devices, then it's unknown what's ok. If your effective hashrate (based on the WU value) is good, I wouldn't worry about the HW error count.

[btc_jumpnrl]

So is under 375-400W per unit the consistent power draw for the S1? I've read several different quotes as the power draw varying from 200-500W and only seen one actual report of 4 units at ~1500 for 375W each a few posts down.

EDIT - Clarified which miner I was refering to

[goxed]

Wrong. 5.5 million diff1 equivalent shares versus 13.5 thousand hardware errors. You're making a classic mistake of comparing the wrong values. His hardware error count is <.3%

Whats the exact relationship to get Diff1 shares? LastDiff*(Accepted+HWerrors)? Thanks



@ 400MHz / 204GH/s the hw error % is 63,253/(63,253 + 3,543,296 + 41,984)% = ~1.7%, which is rather fine for a system operating at its limits Kudos to Bitmain engineers.

[-ck]

Haha, took me a while to get that. So whats the exact relationship to get Diff1 shares? LastDiff*(Accepted+HWerrors)? Thanks

Diff is dynamic on most pools so you can't work out the diff1 equivalent yourself. Leave that to cgminer and it's DiffA + DiffR.

[goxed]

Haha, took me a while to get that. So whats the exact relationship to get Diff1 shares? LastDiff*(Accepted+HWerrors)? Thanks

Diff is dynamic on most pools so you can't work out the diff1 equivalent yourself. Leave that to cgminer and it's DiffA + DiffR.

so DiffA + DiffR is total Diff1 shares right? And should I divide HWerrors/(DiffA+DiffR) to get HW error percentage? Thanks

[-ck]

Haha, took me a while to get that. So whats the exact relationship to get Diff1 shares? LastDiff*(Accepted+HWerrors)? Thanks

Diff is dynamic on most pools so you can't work out the diff1 equivalent yourself. Leave that to cgminer and it's DiffA + DiffR.

so DiffA + DiffR is total Diff1 shares right? And should I divide HWerrors/(DiffA+DiffR) to get HW error percentage? Thanks

HW / (HW + DiffA + DiffR) * 100

[goxed]

Haha, took me a while to get that. So whats the exact relationship to get Diff1 shares? LastDiff*(Accepted+HWerrors)? Thanks

Diff is dynamic on most pools so you can't work out the diff1 equivalent yourself. Leave that to cgminer and it's DiffA + DiffR.

so DiffA + DiffR is total Diff1 shares right? And should I divide HWerrors/(DiffA+DiffR) to get HW error percentage? Thanks

HW / (HW + DiffA + DiffR) * 100

Thanks a lot I will sticky this.

[BITMAIN]

So is under 375-400W per unit the consistent power draw for the S1? I've read several different quotes as the power draw varying from 200-500W and only seen one actual report of 4 units at ~1500 for 375W each a few posts down.

EDIT - Clarified which miner I was refering to

On wall power draw varies depends several things:

• 110v or 220v: Power draw of S1 will be lower if using 220v AC on wall, and higher if using 110v
• PSU efficiency: A golden PSU will have 87%~92% efficiency, while normal 80 plus has 81%. If you use some other PSU without a 80 plus certification, the efficiency would be even lower. http://en.wikipedia.org/wiki/80_Plus
• Voltage drop on wiring: There is 13.4~13.6A current on each blade, we suggest using 3 0.75mm^2 wires or a single wire at 2.5mm^2

According to our measuring, the power draw of the S1 Dual Blades from PSU is 12V*(13.4A*2+0.2A)=324Watt, 0.2A is for the fan when temperature is 25℃.
If you use an 90% efficiency PSU, the power draw from the wall is:324/90%=360watt.

[GandalfG]

Bitmain AntMiner S1 180GH/S Group Buy for 50 Units

BTC 4.25 each to include Free World-Wide Shipping

Ship on December 3, 2013

14 Units In-Stock Remaining...

https://bitcointalk.org/index.php?topic=353650.0

4.5BTC for 180Gh 
5000$/180=27$ for GH/s
In my opinion make no sense spend BTC for hardware at this price point. Better sell BTC, take cash and wait for decrease exchange rate. Than buy  several times more BTC than this machine can generate.

[cxboyminer]

Bitmain AntMiner S1 180GH/S Group Buy for 50 Units

BTC 4.25 each to include Free World-Wide Shipping

Ship on December 3, 2013

14 Units In-Stock Remaining...

https://bitcointalk.org/index.php?topic=353650.0

4.5BTC for 180Gh 
5000$/180=27$ for GH/s
In my opinion make no sense spend BTC for hardware at this price point. Better sell BTC, take cash and wait for decrease exchange rate. Than buy  several times more BTC than this machine can generate.

BTC wont decrease in X rate (like not from 1100 back to 600, maybe 1100 back to 1000). I would rather keep it or buy Antminer