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

[hhumaidan]

As always Bitmain bad quality first batch.

I wonder when is the Avalon 7 is gonna be released.

I will buy the S9 after halving and Bitmain sorts out all issues

[philipma1957]

No Quality Management or just give a fuck about customers

drop to 643 test for about 3 hours
drop to 637 test for about 3 hours
drop to 631 test for about 3 hours
drop to 625 test for about 3 hours
drop to 618 test for about 3 hours
drop to 612 test for about 3 hours
drop to 606 test for about 3 hours
drop to 600 test for about 3 hours
your machine reminds me of my  bad batch 1 s-7

finally got good numbers in the 550 range vs the stock 600

right now  in a hotter space then most I am running 1 s-9 at 587  and 1 s-9 at 625

[d57heinz]

As always Bitmain bad quality first batch.

I wonder when is the Avalon 7 is gonna be released.

I will buy the S9 after halving and Bitmain sorts out all issues

I have to tend to agree here   Be nice for some competition anyway.  To the avg user looks like bitmain sells to us the first test batches until all bugs are worked out.  Leaving that work for us for free no less.  It's like open source only they get all the benefits of the expertise from many of the community.  Then they sell it back to us??  It's a win win for them.  See bitmain has you all between a rock and a hard place.  They know you can't buy hw anywhere else.  They know they can send you junk and there is nothing they have to do about it.  And better yet are you going to waste the time to sue them to make it right.  When you keep rewarding bad behavior then why would they change a thing?

Best regards
D57heinz

[d57heinz]

No Quality Management or just give a fuck about customers

drop to 643 test for about 3 hours
drop to 637 test for about 3 hours
drop to 631 test for about 3 hours
drop to 625 test for about 3 hours
drop to 618 test for about 3 hours
drop to 612 test for about 3 hours
drop to 606 test for about 3 hours
drop to 600 test for about 3 hours
your machine reminds me of my  bad batch 1 s-7

finally got good numbers in the 550 range vs the stock 600

right now  in a hotter space then most I am running 1 s-9 at 587  and 1 s-9 at 625

What I'm not seeing is what are the ambient temps these are runnin in.  With just some high volume fans I'm able to cool gpu fine 100f outside temps.  Just have to move a great volume of air.  With that said. My s4 run at 55c with over clock to 218.75. Need to know if it's even possible to run these at all. And another quick question.  Phil I know you like helping the community and I appreciate your knowledge.  What I'm trying to figure out is why bitmain isn't answering all the questions.  Why do they leave it up to you and a couple others.  Why don't they participate more?

Best regards
D57heinz

[slastar]

Ok, my S9 start working a bit better.

Most important thing was set fan manualy at 90%. After this at 650Mhz get 13TH+ but with quite a lot HW errors (0,0015%). After lower frequency to 625Mh HW drop to 0.0003% and speed stabilized at 13,1TH.
Strange thing is that miner need few restarts to hashig properly, sometimes 1 board hang at 3TH  and dont want speed up.

[philipma1957]

As always Bitmain bad quality first batch.

I wonder when is the Avalon 7 is gonna be released.

I will buy the S9 after halving and Bitmain sorts out all issues

I have to tend to agree here   Be nice for some competition anyway.  To the avg user looks like bitmain sells to us the first test batches until all bugs are worked out.  Leaving that work for us for free no less.  It's like open source only they get all the benefits of the expertise from many of the community.  Then they sell it back to us??  It's a win win for them.  See bitmain has you all between a rock and a hard place.  They know you can't buy hw anywhere else.  They know they can send you junk and there is nothing they have to do about it.  And better yet are you going to waste the time to sue them to make it right.  When you keep rewarding bad behavior then why would they change a thing?

Best regards
D57heinz

Worse yet  they disrespect many good honest people. 

Such as CK ,Kano, myself and sidehack.

But with the solar array  maxing at 11kwatts and only bitmaintech selling .1 or .11 watt gear I purchased 4 units.

Even if we do 50th at 5.5 kwatts  rather then  56 th at 5.5 kwatts.

  I need to have the more efficient gear. as the array's cost is far more then the gear's cost.

[numnutz2009]

hmm which new psu are you talking about?? the miners with the psu mounted on the top? those are only supposed to reach 2.7th/s on the default config. or do you mean the separate 1600w psu (if so those arent "new") they sell? that psu only runs on 220v power i believe so are you sure your using 220v and not 110v?? try using another psu with 4 pci-e power connectors and hook 1 up to the controller and the other 3 to one of the boards. power the miner on with 1 board at a time and see what the hashrate for each board is. this will help you determine which board has issues.

Completely useless suggestions.

First of all, he lives in a country where is 230V electricity .
Second. I'm pretty sure he bought 1600W PSU.
Thirdly. S9 software shows each blade separately. Hashing speed and HV errors.
There is no need to disconnect anything.

I got my first five S9 and one is also "lemon".
Tomorrow brings  courier another 6x S9 . I hope they are better.

first how m i supposed to know where the kid lives?? silly boy think before you speak

second he wasn't clear that's why i asked. there's no reason to assume ne one has 220v power and since idk him personally i didn't want to assume pay attention

third you obviously don't mine much. those that have would know psu issues do come up. for example some times a power supply wont power on one miner but it will work on another one of the same model. i had this happen with 2 s7's i have. the psu would work on every other miner i had except one s7. i tried like 5 s7's and all worked great except for one miner. i had to use a dps-1200fb that i modded as the supply for the controller. the psu that wouldn't power on that single miner was an evga g2 1300w psu so ik the psu was more than beefy enough to power on the controller besides if it wasn't all the other miners wouldn't have powered on either. so before you go runnin ur mouth kid ask. don't say my comments r useless because frankly ur ignorant negative comments don't help at all. i was at least offering a suggestion to him and others like him. what did u provide?? absolutely nothin! so take ur bull elsewhere please or be more polite to your fellow community.

maybe s9's are just arriving dead. i wasn't stupid enough to buy s9's at the current price and risk tens of thousands of dollars until things had a better roi but that doesn't mean i cant try and help people that need it which is more than u can say based on that ignorant reply. 

[numnutz2009]

Ok, my S9 start working a bit better.

Most important thing was set fan manualy at 90%. After this at 650Mhz get 13TH+ but with quite a lot HW errors (0,0015%). After lower frequency to 625Mh HW drop to 0.0003% and speed stabilized at 13,1TH.
Strange thing is that miner need few restarts to hashig properly, sometimes 1 board hang at 3TH  and dont want speed up.

those chips temps are wayyyy too high my friend. your running some dangerous numbers. they should never go much higher than 70 degrees c at the chip and each day it runs like that it loses a little more life until it finally dies. if posible i would try fixing this sooner rather than later. i dont want your hardware throwin out that magic smoke.

[numnutz2009]

What a dissapointing miner. Got 5 pcs S9`s and only 1 is working 100% even after firmware upgrade.
1 miner is running only 2 th on 1 board another miner is only running 2.4 TH on all 3 boards and 2 are completely dead. I bought even the new Bitmain PSU with the miners...
Holy moses im dissapointed.

wow that sucks bad.


my two

1 does freq 625 good temps.
1 does freq 587 good temps.

I do firmware after all other methods are tried.

I would have down clcoked the 2th at

 freq 650
 then 625
 then 600
then 575
 then 550
 then 525
then 500 if that failed

I would have pulled the psu from the best miner running close to spec and put it on the 2th 1 board miner. and tried freq 500 first.

if it worked I would know the boards are okay and the psu was shit.
if it did not work I would know the miner has an issue.


@biodom

1 very stable at freq 625 fans at 90%  about 13250
1 very stable at freq 587 fans at 90%  about 12750

they are good from 75 f to 90 f

Im planning to try out lower freq on 2 of them today even they are all running at very cool temperatures. 80C on Chips.
I just do not understand why 2 are just dead….  No light no nothing. Tried 3 different Bitmain 1600 psu's on them and they just do not react at all…

but do you have another non bitmain psu that you can connect up to the miners or just to the controller to see if they power on. im not saying their monster psu's are bad or anything i just don't have 220v so i never used them but maybe the controller doesn't like the bitmain psu sorta like one of my s7's didn't like my evga g2 1300w psu while all my other miners worked fine with it. its just an option.

[Puffy23]

Ok, my S9 start working a bit better.

Most important thing was set fan manualy at 90%. After this at 650Mhz get 13TH+ but with quite a lot HW errors (0,0015%). After lower frequency to 625Mh HW drop to 0.0003% and speed stabilized at 13,1TH.
Strange thing is that miner need few restarts to hashig properly, sometimes 1 board hang at 3TH  and dont want speed up.

those chips temps are wayyyy too high my friend. your running some dangerous numbers. they should never go much higher than 70 degrees c at the chip and each day it runs like that it loses a little more life until it finally dies. if posible i would try fixing this sooner rather than later. i dont want your hardware throwin out that magic smoke.

It would be nice for Bitmain to weigh in on this discussion.  I know running the PCB past 70C can increase life-cycle maintenance or decrease effective life (I'm sure most of us have dealt with failed heatsink adhesive), but what are the chips comfortable running at?  I've heard of chips running comfortably at 130C and some that don't like anything over 90C.

I run my S7s at 55C at the board.  Anyone know what the likely chip temperatures are based on board temps?

[Biodom]

it would be interesting to tally up S9 B1 performance so far. There seem to be some complains, but let's put some real numbers on this.
Please respond (if you want) with four  parameters. If machine is DOA, obviously zeroes for speed.

1. Th at start, for example 1hr without mhz adjustment (if possible)  
2. Stable Th
3. stable mhz
4. ambient temp

So far (of what was posted in bits and pieces):

[Tupsu]

hmm which new psu are you talking about?? the miners with the psu mounted on the top? those are only supposed to reach 2.7th/s on the default config. or do you mean the separate 1600w psu (if so those arent "new") they sell? that psu only runs on 220v power i believe so are you sure your using 220v and not 110v?? try using another psu with 4 pci-e power connectors and hook 1 up to the controller and the other 3 to one of the boards. power the miner on with 1 board at a time and see what the hashrate for each board is. this will help you determine which board has issues.

Completely useless suggestions.

First of all, he lives in a country where is 230V electricity .
Second. I'm pretty sure he bought 1600W PSU.
Thirdly. S9 software shows each blade separately. Hashing speed and HV errors.
There is no need to disconnect anything.

I got my first five S9 and one is also "lemon".
Tomorrow brings  courier another 6x S9 . I hope they are better.

first how m i supposed to know where the kid lives?? silly boy think before you speak

second he wasn't clear that's why i asked. there's no reason to assume ne one has 220v power and since idk him personally i didn't want to assume pay attention

third you obviously don't mine much. those that have would know psu issues do come up. for example some times a power supply wont power on one miner but it will work on another one of the same model. i had this happen with 2 s7's i have. the psu would work on every other miner i had except one s7. i tried like 5 s7's and all worked great except for one miner. i had to use a dps-1200fb that i modded as the supply for the controller. the psu that wouldn't power on that single miner was an evga g2 1300w psu so ik the psu was more than beefy enough to power on the controller besides if it wasn't all the other miners wouldn't have powered on either. so before you go runnin ur mouth kid ask. don't say my comments r useless because frankly ur ignorant negative comments don't help at all. i was at least offering a suggestion to him and others like him. what did u provide?? absolutely nothin! so take ur bull elsewhere please or be more polite to your fellow community.

maybe s9's are just arriving dead. i wasn't stupid enough to buy s9's at the current price and risk tens of thousands of dollars until things had a better roi but that doesn't mean i cant try and help people that need it which is more than u can say based on that ignorant reply.  

If you're smart enough not to buy the S9, why are you commenting on things that you do not have and of which you know nothing about ?  Not smart enough for this ?
I had 43x S7, burnt down 3 blade during the warranty period and 5 blade after  the warranty period.
 I have never seen a problem , that PSU working with one S7, but will not work with another S7. Then, the power supply is simply too weak. But 1600W Bitmain PSU  can not be too weak for S7 or S9

I have now 11 xS7 with 2 x EVGA G2 1600W , 1 x Bitmain 1600W PSU, with 2 x IBM 2000W single PSU and 3x IBM 2000W double PSU .
The best mining results gives me 1600W EVGA  and APW3-12-1600-B2 (old version , with 12x PCI-E cables)

[slastar]

Ok, my S9 start working a bit better.
https://s32.postimg.org/53zy2xn91/s9_625_Mhz.png
Most important thing was set fan manualy at 90%. After this at 650Mhz get 13TH+ but with quite a lot HW errors (0,0015%). After lower frequency to 625Mh HW drop to 0.0003% and speed stabilized at 13,1TH.
Strange thing is that miner need few restarts to hashig properly, sometimes 1 board hang at 3TH  and dont want speed up.

those chips temps are wayyyy too high my friend. your running some dangerous numbers. they should never go much higher than 70 degrees c at the chip and each day it runs like that it loses a little more life until it finally dies. if posible i would try fixing this sooner rather than later. i dont want your hardware throwin out that magic smoke.

No they are not. For GPU yes, but not for ASICs.

[clgrissom3]

Being able to see the individual blades' performance opens up the possibility of combining good or bad performing blades into a single unit and then clocking the unit appropriately for the blades (presumably voiding the warranty in the process though).

@yxt have you tried downclocking at all to see what happens to the blades that are not performing as well?



Also out of curiosity, has anyone from batch 2 or 3 received a shipping notification?  I am curious if batch 3 really is going to ship before batch 2.

My batch 3 is scheduled to arrive Monday the 20th...

[Cassey]

Buddy, Slush's has the BEST dashboard, but CKPool has the BEST profits. 

Can we have a non-emotional, fact based, discussion about that?  I am open to switching.

Slush is running around 62PH, it has a 2% fee and using a scoring system that seems fair (rewards consistent participation).  Transaction fees are shared.

CKPool is running around 36PH, has a 0.9% fee and uses PPLNSG payouts.  Transaction fees are also shared.

Seems to me, for someone that is running for months, PPLNSG and Slushes scoring system are similar in effect, both avoid block hoppers (are those still a problem?).

Variance based on pool size is not a big deal for either:  both are large enough that a daily payout is reasonable (my personal threshold).

Seems Slush wouldn't make for a very good backup pool, since short term access results in a lower score.  Seems to me that CKPool may be ok for that, although the payouts would be delayed by a day or so.

Bottom line, the fees are the major difference?

BTW - I have CKPool as one of my backup pools as of last night, but all three of my miners are showing the pool as dead.  I'm using a named account.  The doc indicates the worker name will be auto-generated upon first contact, yet I'll never have a first contact if the pool doesn't show up as alive...

[Puffy23]

Almost at 2 hours now, web page indicateds 13,544.19 GH/S(RT) and 13,6954.32(avg) - just added the "Lowes sound mod" and noticing the temps go up a few degrees - now at 65-67C.

Of interest, Slushpool is crediting my S9 with 14.53TH/s over the past hour:

My units are in a basement "machine room", so thinking of removing the grills to see if that helps.  On my S7s, I found partially blocking the intake worked, but raised temps.  Using a block of 2x3 lumber BEHIND the unit brought the sound under control.  May play a bit more before disabling the safety grills.

Buddy, Slush's has the BEST dashboard, but CKPool has the BEST profits. 

I use Slush because I can have an active dashboard on every single electronic device I own.  I notice within minutes when one of my miners is offline or hashing at unusually low rates.  The increased % of uptime is worth more to me than the higher % of blocks found.

My units are in a remote location on a TWC business class connection and I still live in fear that they will burn up on an internet outage.  I've had several that I noticed immediately and by the time I arrived it still smelled a bit toasty and the metal racks were almost searing hot.  No other pool offers the kind of access to the dashboard (from my experience) that allows me to respond almost immediately.

[SaBit]

Blade 4 seems to have 10x the HW errors   -  Now hashing at 13.444Thz  -- Ambient temp is below 20deg C





Oh  and numnutz2009  your name is Fitting -- Just replace the first  n with a D

[citronick]

it would be interesting to tally up S9 B1 performance so far. There seem to be some complains, but let's put some real numbers on this.
Please respond (if you want) with four  parameters. If machine is DOA, obviously zeroes for speed.

1. Th at start, for example 1hr without mhz adjustment (if possible)  
2. Stable Th
3. stable mhz
4. ambient temp

So far (of what was posted in bits and pieces):

My S9 stabilized to a level I am comfortable with at 650mhz this time hashing 13.7 THs (showing 14.74THs at Kano pool) but fans is still almost full speed at 95%, no woo-woo sound at least. PCB 57-62c, CHIP 86-95c.

After the firmware upgrade, things seemed to have done the trick - but the 3rd blade is producing high HW error rate - in the hundreds. After 10 hours or so, HW error count from balde 1 to 3 are, 19, 32, 455! This bit is still bothering me.

[jtoomim]

it would be interesting to tally up S9 B1 performance so far. There seem to be some complains, but let's put some real numbers on this.
Please respond (if you want) with four  parameters. If machine is DOA, obviously zeroes for speed.

1. Th at start, for example 1hr without mhz adjustment (if possible)  
2. Stable Th
3. stable mhz
4. ambient temp

Here's some more precise numbers. I only own Machine 1, so that's the only one I've tried to overclock so far. Still testing its limits.

Machine 1:
1. 13.7 TH/s
2. 14.2 TH/s
3. 668 MHz
4. 10°C - 20°C
5. 55°C max Temp(PCB)
6. 85°C max Temp(Chip)

Machine 2 ("m5"):
1. 13.7 TH/s
2. 13.7 TH/s
3. 650 MHz
4. 10°C - 20°C
5. 55°C max Temp(PCB)
6. 86°C max Temp(Chip)

Machine 3 ("m6"):
1. 8.0 TH/s
2. 11-12 TH/s
3. 575 MHz
4. 10°C - 20°C
5. 55°C max Temp(PCB)
6. 86°C max Temp(Chip)

Machine 4 ("m7"):
1. 13.8 TH/s
2. 13.8 TH/s
3. 650 MHz
4. 10°C - 20°C
5. 56°C max Temp(PCB)
6. 87°C max Temp(Chip)

Machine 5 ("m8"):
1. 7.5 TH/s
2. 11.5-12.5 TH/s
3. 575 MHz
4. 10°C - 20°C
5. 53°C max Temp(PCB)
6. 86°C max Temp(Chip)

[philipma1957]

Buddy, Slush's has the BEST dashboard, but CKPool has the BEST profits. 

Can we have a non-emotional, fact based, discussion about that?  I am open to switching.

Slush is running around 62PH, it has a 2% fee and using a scoring system that seems fair (rewards consistent participation).  Transaction fees are shared.

CKPool is running around 36PH, has a 0.9% fee and uses PPLNSG payouts.  Transaction fees are also shared.

Seems to me, for someone that is running for months, PPLNSG and Slushes scoring system are similar in effect, both avoid block hoppers (are those still a problem?).

Variance based on pool size is not a big deal for either:  both are large enough that a daily payout is reasonable (my personal threshold).

Seems Slush wouldn't make for a very good backup pool, since short term access results in a lower score.  Seems to me that CKPool may be ok for that, although the payouts would be delayed by a day or so.

Bottom line, the fees are the major difference?

BTW - I have CKPool as one of my backup pools as of last night, but all three of my miners are showing the pool as dead.  I'm using a named account.  The doc indicates the worker name will be auto-generated upon first contact, yet I'll never have a first contact if the pool doesn't show up as alive...

Slush is bad if you have a blackout  you lose your work.  I would not use them for this one reason. Forget any other reason then this one.

also along the same lines if you do a big rental  on nicehash and the rental drops off you lose your work.

I don't mine anywhere that loses my work  in under 12 hours.  To me the 5n feature of kano is a big reason I mine with them.