Bitmain's Released Antminer S9, World's First 16nm Miner Ready to Order

[E]

how this (66min for a block) is even theoretically possible? It seems to go way over five sigma.
does it mean that a lot of hashing power is fake a.k.a INTENTIONALLY setup to NOT find blocks? is this even possible?

Surprisingly 66min is just under 3sigma if the network is running at 10min/block:

http://www.wolframalpha.com/input/?i=CDF%5BExponentialDistribution%5B1%2F10%5D,+66%5D

0.99864

thanks for really running the numbers-I just guesstimated; it is surprising, indeed.

Yeah, it's counter to my intuition for sure. Turns out 5 sigma would be 144 minutes:

Code:

Solve[CDF[ExponentialDistribution[1/10], x]==(1-5.722*10^-7), x]
{{x->143.738}}

http://www.wolframalpha.com/input/?i=Solve%5BCDF%5BExponentialDistribution%5B1%2F10%5D,+x%5D%3D(1-5.722*10%5E-7),+x%5D

[1714920785]

1714920785

[1714920785]

1714920785

[citronick]

S9 review:

At 650 mhz default frequency, temps at chip reached over 100c when it started mining, is this normal? But PCB temps hovered around 60-64c. So, I put fans at 100% but chips temp still within the 99c-101c range.

Adjusted the freq to 625 mhz, and cranked up fans 100%, now chip temps 88-94c, PCB 56-62c.

I noticed that the 3rd board run hotter than the rest.

Is the chip temp normal at that high temps when mining?

Appreciate any feedback - thanks!

[E]

S9 review:

At 650 mhz default frequency, temps at chip reached over 100c when it started mining, is this normal? But PCB temps hovered around 60-64c. So, I put fans at 100% but chips temp still within the 99c-101c range.

Adjusted the freq to 625 mhz, and cranked up fans 100%, now chip temps 88-94c, PCB 56-62c.

I noticed that the 3rd board run hotter than the rest.

Is the chip temp normal at that high temps when mining?

Appreciate any feedback - thanks!

That's in the ballpark of what I'm seeing at 650Mhz, 100% fan, 20C intake. The chip temps aren't too worrying - they probably aren't calibrated and a on-die/Tjunction specification of 100C or so is believable.

The 1st and 3rd boards will tend run hotter than the center one due to the airflow design, with the 3rd the hottest because of the asymmetry of which side of the board is exposed to more airflow.

[Biodom]

S9 review:

At 650 mhz default frequency, temps at chip reached over 100c when it started mining, is this normal? But PCB temps hovered around 60-64c. So, I put fans at 100% but chips temp still within the 99c-101c range.

Adjusted the freq to 625 mhz, and cranked up fans 100%, now chip temps 88-94c, PCB 56-62c.

I noticed that the 3rd board run hotter than the rest.

Is the chip temp normal at that high temps when mining?

Appreciate any feedback - thanks!

that pict on the left is surely distracting...lol

SP20 chips , which are probably the same type of silicon, were fine until 115C, showed warning at 120C and were program to throttle at 125C, from which i surmise that 100C chip temp is just fine and dandy.

[NotFuzzyWarm]

S9 review:

At 650 mhz default frequency, temps at chip reached over 100c when it started mining, is this normal? But PCB temps hovered around 60-64c. So, I put fans at 100% but chips temp still within the 99c-101c range.

Adjusted the freq to 625 mhz, and cranked up fans 100%, now chip temps 88-94c, PCB 56-62c.

I noticed that the 3rd board run hotter than the rest.

Is the chip temp normal at that high temps when mining?

Appreciate any feedback - thanks!

ref https://bitcointalk.org/index.php?topic=1512186.msg15212835#msg15212835
I've dropped my b1 down to 600MHz to keep temp of the hot board under 100C with the ambient in my basement being around 85F.

[yslyung]

S9 review:

At 650 mhz default frequency, temps at chip reached over 100c when it started mining, is this normal? But PCB temps hovered around 60-64c. So, I put fans at 100% but chips temp still within the 99c-101c range.

Adjusted the freq to 625 mhz, and cranked up fans 100%, now chip temps 88-94c, PCB 56-62c.

I noticed that the 3rd board run hotter than the rest.

Is the chip temp normal at that high temps when mining?

Appreciate any feedback - thanks!

https://bitcointalk.org/index.php?topic=1512186.msg15226593#msg15226593

i posted something, could be useful

[NotFuzzyWarm]

S9 review:

At 650 mhz default frequency, temps at chip reached over 100c when it started mining, is this normal? But PCB temps hovered around 60-64c. So, I put fans at 100% but chips temp still within the 99c-101c range.

Adjusted the freq to 625 mhz, and cranked up fans 100%, now chip temps 88-94c, PCB 56-62c.

I noticed that the 3rd board run hotter than the rest.

Is the chip temp normal at that high temps when mining?

Appreciate any feedback - thanks!

https://bitcointalk.org/index.php?topic=1512186.msg15226593#msg15226593

i posted something, could be useful

Missed that somehow. Good info to work with.
Still keeping at 600 though as that hottest board is also what generates the most HW errors. Below 100C it's less than a total of 1 HW error every 500 accepted shares with almost all coming from that 1 card. At higher freq and temp 102-103C the errors from that card alone were 1 error every 75 accepted shares. yeesh...

[citronick]

Also note that it has the same # of chips as b1 (189) but runs at a lower clock of 600MHz. Looks like Bitmain decided that the b1's were actually now what they consider as OC'd... Their way of addressing the high chip temps?

Und ja, I ordered 1 of the batch-3's.

Oh, my b1 miner seems to have settled down in it's new home and the overnight avg hash rate as reported by Kano is 13.5THs and checking just now is holding at 13.7 average on the gui. Much better than the barely 13THs it started out at.

I did your fix using vent filters on the intake side.


http://www.lowes.com/pd/WEB-Web-Vent-Register-Filter/3122205

duct tape.


http://www.lowes.com/pd/3M-1-88-in-x-135-ft-Gray-Duct-Tape/3505480


99% of air gets through .

quite a lot of dust is stopped.

and no train whistle.

under 10 bucks and will do 12 s-9's


625 freq is giving me about  26 to 27 th.

temps are 65-90

earnings are 0.03175 so far.

Hi Philip, can you show the pix for - vent filters on the intake side?
3 of my S7s suffered because of dust problems - hate to have that on the S9. Thanks

I am going back to solar array in a few days but I can do a mock up for you.

shot before filter


this is mock up.

 red electrical  tape not duct tape

white piece of felt not 4 inch  by 10 inch  vent filter

and the fan is wrong a silverstone 141  not a bitmaintech tech.
of course a card board box the fan is on is not the s-9
but this is the general idea  in real life the filter is long like the fake photo but pretty much is perfect fit height wise


top view



forward view

Thanks Philip, I get it - good mockup.

I will try it but I will need to monitor the temps.

I dont have the woo-woo sound so I have 1 less problem to worry about.

But the dust issue is a serious as it got my S7s to lose hash.

I got to go the DIY shop to look for these vent filters, so while searching the house for something similar....

I thought a "pantyhose" to replace the vent filter will be a good idea - at least to address the dust issue.

I took some of my wife's pantyhose to test... I am sure she will not mind... looks promising.

But yes, I still need a proper vent filter / muffler + pantyhose combination, so that I can further reduce the sound and reduce dust.

[E]

So they made sure to dual-FET both high and low sides of the buck switching, which is good considering the high currents they're seeing. The FET section is also heatsinked, I like to see that.

This is good to hear. The 2/1 FET setup on the S7 buck really caught my eye when I first pulled an S7 board. They clearly thought about it since the pads were there.

Have you measured chain voltage yet?

[kodokbuduk]

why you guys worry about dust on the machines? I've got dust on my machines and simply take out the fans, use brush and vacuum the dust that block the heat sink and put it back both fan once a month/two

[sidehack]

No, I didn't want to mess with things that much since it was a loaner. If I had one of my own It'd have been completely torn down inside an hour. I didn't do anything that required applying power or actual disassembly.

[citronick]

Also note that it has the same # of chips as b1 (189) but runs at a lower clock of 600MHz. Looks like Bitmain decided that the b1's were actually now what they consider as OC'd... Their way of addressing the high chip temps?

Und ja, I ordered 1 of the batch-3's.

Oh, my b1 miner seems to have settled down in it's new home and the overnight avg hash rate as reported by Kano is 13.5THs and checking just now is holding at 13.7 average on the gui. Much better than the barely 13THs it started out at.

I did your fix using vent filters on the intake side.
<snip>

Thanks Philip, I get it - good mockup.

I will try it but I will need to monitor the temps.

I dont have the woo-woo sound so I have 1 less problem to worry about.

But the dust issue is a serious as it got my S7s to lose hash.

I got to go the DIY shop to look for these vent filters, so while searching the house for something similar....

I thought a "pantyhose" to replace the vent filter will be a good idea - at least to address the dust issue.

I took some of my wife's pantyhose to test... I am sure she will not mind... looks promising.

But yes, I still need a proper vent filter / muffler + pantyhose combination, so that I can further reduce the sound and reduce dust.

[citronick]

why you guys worry about dust on the machines? I've got dust on my machines and simply take out the fans, use brush and vacuum the dust that block the heat sink and put it back both fan once a month/two

what about the dust or grit that get into the PCB itself?

[kodokbuduk]

why you guys worry about dust on the machines? I've got dust on my machines and simply take out the fans, use brush and vacuum the dust that block the heat sink and put it back both fan once a month/two

what about the dust or grit that get into the PCB itself?

i didnt touch that, just leave it..not much anyway. If you put something/filter in front of the fan, then it is same having dust blocking the airflow..

[kodokbuduk]

Also that whistle sound will be gone by time due to dust blocking the airflow..once i clean it, then it happen again..so just let it be

[citronick]

S9 review:

At 650 mhz default frequency, temps at chip reached over 100c when it started mining, is this normal? But PCB temps hovered around 60-64c. So, I put fans at 100% but chips temp still within the 99c-101c range.

Adjusted the freq to 625 mhz, and cranked up fans 100%, now chip temps 88-94c, PCB 56-62c.

I noticed that the 3rd board run hotter than the rest.

Is the chip temp normal at that high temps when mining?

Appreciate any feedback - thanks!

that pict on the left is surely distracting...lol

SP20 chips , which are probably the same type of silicon, were fine until 115C, showed warning at 120C and were program to throttle at 125C, from which i surmise that 100C chip temp is just fine and dandy.

LOL - yes, a few members have given me some feedback on that pix...

meanwhile thanks for the temps feedback - the unit is at 625mhz, PCB low 60s, chip 88-95s - so i am happy camper now.
I will crank it up tonight when its cooler, so that I could see this guy run at 650mhz or more.... anyone tried 700mhz?

[citronick]

Also that whistle sound will be gone by time due to dust blocking the airflow..once i clean it, then it happen again..so just let it be

ok good point - regular maintenance is required anyways.
at least the pantyhose will do for now, once dust builds up, i can always use the other "leg" of the pantyhose. 

[kodokbuduk]

yea, and last time I've experience one of the fan suddenly is not working (broken into pieces, dont know why), then the miner just stop, beeping sound.. I replaced it (ordered from bitmain part site, 6 days to arrive), and back to work like a charm..So this is good, i know this miner has prevention of overheating. Now i got spare fans, so if this happen again, i got spare, no need to wait 6 days to order.

[NotFuzzyWarm]

Also that whistle sound will be gone by time due to dust blocking the airflow..once i clean it, then it happen again..so just let it be

ok good point - regular maintenance is required anyways.
at least the pantyhose will do for now, once dust builds up, i can always use the other "leg" of the pantyhose.  

Problem with letting the dust bunnies start living in the miners is not just one of them restricting airflow. Surface contamination of the heatsink fins by fine dust rather dramatically begins to impact heat xfr efficiency before airflow throttling becomes a concern. Miners like all the cooling they can get. Yes, they can tolerate what I would call heavy abuse and possibly/probably do it for a rather long time.

But still, when so easy to kill 2 birds with 1 stone...

[clgrissom3]

I couldn't stand it anymore...I broke down and pulled the trigger for an S9 tonight...   I will need a couple of these to keep my host fee average down after the halving.  

I might be the first person to complete an S9 purchase in less than an hour!  Crazy!  As soon as my purchase was marked as paid Batch 3 changed to Sold Out...