Topic: Bitmain's Released Antminer S9, World's First 16nm Miner Ready to Order - page 249. (Read 531168 times)

what made you think that i am homeless,my parents are rich if you want to know,that does not mean i do own a house or i am earning enough to buy a house. I am just an up and coming guy who is trying to have an aspiring career and that is it. Started my career with 250$ a month and i am just moving up, i am from india and things are like this over here.

Chrome on my mobile.

What browser did you use ? I tried Chrome and firefox with same probem ?

I haven't had any problem ordering. I ordered 12 last week with no problem.

It has no problem cooling because your 3KW of PSU are burning about 300W of heat in themselves (assuming ~90% efficiency) so the power density is a fair bit less. The heat-generating parts in a PSU are probably also either solid metal or a big beefy silicon rated for 150C max internal temperature, rather than high-speed data paths on the most ornately etched crystal in the world, making the hardware a bit less finicky overall.

But yeah, one of the S9 boards that failed within 2 days in my hosting I don't think ever saw over 90F ambient temps, and were probably between 75F and 85F for most of that time. It also started to hash but died out quickly when the frequency was turned below 300MHz. Kinda strange.

@numnutz!

Well to explain myself.  I run them in a barn. Outside temp wasn't 120. But inside temp of barn was. Ask Phil how hot it is where he runs his miners;)  Possibly higher.  Yes you are correct on the miners and the heat sink.  Can you tell me why bitmain is going backwards with reliability and forwards with efficiency.  Let's make the first ASIC with a great heat sink but we all know that it's not here to stay.  Then let's move to today. New ASIC with most efficient chip and the worst track record.  Kiddos as you put it so kindly must have been who is assembling these.  Cheap pcb cheap heat sinks. Fires burnt boards galore.  Warranty is next to the shit on the bottom of my shoe. And now as a big fuck you they release the antrouter r 4 lmao.   I'm waiting on Avalon to deliver and will continue to buy gpu.  Good day to you.  

EDIT   Lucky for me if you want to push this much further i still own s1 s2 s3 s4 c1's  would you  like me to run them in a closed env to see what the thermal breakdown is.. Think this might just make for a great youtube journey for me.  and some lessons for the community on how the old miner worked so much better!!!  and dont worry ill put enough to equal the same amount of watts.  again you basically explained what i needed to know .  when they burn up in datacenters then how do you expect them to be ran at home?  ill do some digging and see if i can find max temps.. Im really just over this.  

Btw I'm not going anywhere:)

BR


S3+ Specifications:

 

Hash Rate: 453 GH/s±5% with great OC potential
Power Consumption: 355 Watt at the wall
Power Efficiency: 0.78J/GH on wall
Power Supply: 12V DC
Size:  331 mm x 137 mm x 160 mm
Fans: Two 14038 fans mounted on both front and back ends
Operating Temperature: 0 °C to 35 °C
Complied with: FCC / CE
Network Connection: Ethernet

just to let you know heehhe.. look at line 7

S9 Specifications:

1. Hash Rate: 11.85TH/s ±5%

2. Power Consumption: 1172W + 12% (at the wall, with APW3 ,93% efficiency, 25C ambient temp)

3. Power Efficiency: 0.1J/GH + 12%(at the wall, with APW3 93% efficiency, 25°C ambient temp)

4. Rated Voltage: 11.60 ~13.00V

5. Chip quantity per unit: 189x BM1387

6. Dimensions: 350mm(L)*135mm(W)*158mm(H)

7. Cooling: 2x 12038 fan

8. Operating Temperature: 0 °C to 40 °C

9. Network Connection: Ethernet

10. Default Frequency: 550M

and one just has to assume that if you are running in those conditions and as long as the box is checked in the miner for it not to exceed 80c(i know its not stated in manual but 80c is max) then it should be safe to run up to 40c conditions.. And yea i took some risk. but it worked great for me.  .. i didnt quite understand your question of max temp that is why i put the op temps of s3  stating that it at least gave a range it is SAFELY ran in.  anyway.. i dont understand all things and ill be the first to admit it.. But your way of making me understand sure wont work in any classroom environment.  im reasonable but i also wont be fucked around either.

Also can you tell me why i can smash 3kw of psu into a box 1/3 size of s9 and it has no problems whatsoever cooling?  maybe bitmain needs to go back to the drawing board?

Yes, and now they agreed to get them even without discussing with them....and without opening a Ticket. I do not know though thats because of the volumes I buy, I really hope not, since everyone should have the same service level. Secondly, I need to buy more and Avalon is just too slow at the moment.
Still waiting though, but im affriad that I wil need to dish Avalon if they keep putting out miners who demands PI as controller, its a freaking mess to operate a farm with all these PI`s.

Did you try and get warranty repair on your failed S9 board?  Just curious.  Surprised to see you buying more.


I have one board on one miner that occasionally shows x's at stock freq but I have it downclocked and it doesn't happen.  The board in question is clearly problematic, at stock it sometimes only hit about 60% of expected hashrate and never went any higher.  It would take me several reboots to get it to come up higher but even then it wouldn't hit the numbers of the other two boards, and it would pop up occasional x's as mentioned.  When it was doing that the x's would come and go although usually on the same 4 or 5 chips.  So yeah, the whole thing is downclocked and it doesn't happen.

IS the shop down.
I have tried now for days to make a new order of 10 x S9, but at checkout it stops when getting to shipping destination ??
Anyone else having the same problem ?

https://gyazo.com/8e37d262da3d55b6908e9a7497b7e416

one of my S9 is showing 1 X in chips.
Tried rebooting it still showing an X
Hopefully it won't start showing more and more Xes

I'm living on less than one S9 worth of money in a given month, and it's only that much because I have an old credit card and student loans to pay down. Actual living expenses (rent, utilities, groceries) I could probably get by on about half an S9 per month, and that's right here in the US. Depending how and where you live, it's not even hard. I had a neighbor for a while with a wife and baby doing fine on a household income of under $300 per week.

The last miner I bought new was an S3, near the end of their run when prices were like $280. S9 prices would have to be about half where they are now for me to think about buying, and I've probably got cheaper power than most of y'all. And even then I probably wouldn't buy one because Bitmain hasn't done much in a really long time to demonstrate a desire to not actively screw people over.

Money is relative.  I live in NJ in a 375,000 usd home depending on what city in NJ that price can be a small home in really nice town or a medium home in a medium priced town .

Or a pretty big house in a poor town.

You can range in size from  1200 square ft .  To 3750 square foot.

You may even get a big lot for that price say four acres.

Or in some towns a small lot 4000 square ft.

From country #100 downward, average monthly salary is less than $633.
India- $452/mo, China $732, Russia $686. Things cost less in these three, I imagine.
You can survive even in US on this money (if you don't have to pay rent or with very low rent in some areas).
http://www.nationmaster.com/country-info/stats/Cost-of-living/Average-monthly-disposable-salary/After-tax
Also, the last country has $25, but i don't think that they are necessarily starving. Everything just cost less to much less.

How do you survive? are you homeless?

Never thought you need to spend this much amount of money to mine. Are people here really rich or what,spending this much dime for a machine. My three month salary for one machine  carry on folks.

Definite JA on the switching noise issue. I've brought up the point in several other threads regarding PDN considerations. It's not so much clk speed that hurts as it is circuit current rise/fall times that can cause very localized havoc. For one ref, http://powerelectronics.com/power-electronics-systems/five-things-every-engineer-should-know-about-pdn

Considering that as part of the packaging into an actual chip the dies are mounted to an interposer to connect the outside world, that would be a perfect place for the temp sense diode. Giver the power being handled prolly at worst alumina and preferably with AlN as the interposer substrate, either way if there is point on it to locate the temp diode -- close enough for 'die' temp in my book.

On Bitmain and temp modeling... Probably done, probably pretty basic. Think the best we can hope they do is accelerated life testing at a couple different high to very high temps...

On your homework problem: heh heh heh... Things truly get astounding even for me at the extremes of current physical material science. My current playtoy, er, um, testbed, at work is a Trumpf Ultrafast laser. To cut to the chase, one setting provides 85uJ per 1ps wide pulse firing every 5us. That translates to a peak pulse power of 85MW. Now focus that down to a 25um dia spot and calc the power density in W/sq cm that the target sees (on that one tiny spot) during that 1ps. Ballpark hint: It's in the tera-watts... Good freaking lord.

QC- I had one S7 miner with 575 sticker on top of 550 sticker, i was charged for 575 miner, it worked as 575 for a month or two (barely), then had to be put on 550, where it actually belong. Such corners cutting is typical here. I did not made the fuss because it all averages, in my opinion. i got one other that was better than expected.

In my honest experience to be working for a company that build electronics devices. If the a board failed the QC then they tried it again and it worked, it should never been sold to anyone. You might have a solder issue on the board. (It can be a defective parts or whatever possible)

It might be working cold ... or hot or in between but it doesn't work all the time. You could bend the PCB a bit and it would be working again ... but for how long ? This is bad QC.

When it does happen for us, this is a major issue and we take it seriously has many factors can influence the exact problem. We stop the production if we have to... until we are 100% sure the cause of the issue is fixed.

Anyway this is just my 2 cents

Unfortunately it isn't that simple. There are two issues:

1) lack of calibration. Performing a proper calibration of the internal temperature sensor would be probably prohibitively expensive. Uncalibrated sensor will have good differential accuracy (is the temp higher or lower?) but bad absolute accuracy (is it 100C or 120C?).
2) internal switching noise. The mining chips have super-high levels of internal simultaneous switching noise, beyond what is tolerated by the standard library IP blocks.

Spondoolies tried to work around it by configuring that temperature sensor cell to provide intentionally coarse measurements; IIRC the step was 5C.

The other way to work around those limitations is to use external measurement chip (like LM75) with external sensing diode that is internal to the mining chip. This setup is much less sensitive to the manufacturing process variations.

The determination of true maximum operating temperature would be too time consuming and too expensive. There published values are just very approximate goals that are used as an input to the electromigration models in the CAD software. And again the normal operating points of the mining chip are outside of the ranges where those models have any accuracy. So it is back to keeping the fingers crossed. There's no time for the proper qualification procedures and fault analysis.

In addition to the above the maximum safe operating temperature would be dependent on the core supply voltage. For higher voltages the temperatures would be lower. My textbooks are in storage, I can't look them up now. But anyone interested could dig around the web and try to find some school problems like: the capacitor has plates at a distance of 10nm, the voltage difference between plates is 1V. What is the equivalent pressure in PSI required to keep those plates apart? The results are frighteningly high. Modern processed utilize pre-stressed silicon to partially deal with those electrostatic forces.

Losing 1 bee wont make or break ne thing.....its not like its the queen after all lol.