Author

Topic: ANTMINER S7 is available at bitmaintech.com with 4.86TH/s, 0.25J/GH - page 139. (Read 527798 times)

hero member
Activity: 840
Merit: 1000
B6 backside photo https://imgur.com/sIYNhVX

Ok, so, the "30C" resistor is the 20k reference one.

and one of the "123" resistor allows you to output 10.6v.
divided by 15 chips in a string = 0.71v at the chips

1.1k resistor in place of the 1.3k will give 0.77v at the chips
1.3k resistor will give 0.65v at the chips
18k resistor in place of the 20k should also work for 0.64v at the chips
22k resistor in place of the 20k will give 0.77v at the chips

Edit, I'm aging, took me 20 minutes to figure the over/undervolt.


Does that mean i can simply slap some graphite on 30C (20k resistor) to undervolt the Antminer S7? That would be simple and interesting for the long run.

yes, it should work, but since the resistor in on board, the value you can read will be different from 20k.
So as for the good old antS1, you will need to do it by trials and errors until someone check the original value with a 20k resistor on board, and the target value with a 18k resistor (or 16k for 0.57v at the chips)
legendary
Activity: 1302
Merit: 1068
B6 backside photo https://imgur.com/sIYNhVX

Ok, so, the "30C" resistor is the 20k reference one.

and one of the "123" resistor allows you to output 10.6v.
divided by 15 chips in a string = 0.71v at the chips

1.1k resistor in place of the 1.3k will give 0.77v at the chips
1.3k resistor will give 0.65v at the chips
18k resistor in place of the 20k should also work for 0.64v at the chips
22k resistor in place of the 20k will give 0.77v at the chips

Edit, I'm aging, took me 20 minutes to figure the over/undervolt.


Does that mean i can simply slap some graphite on 30C (20k resistor) to undervolt the Antminer S7? That would be simple and interesting for the long run.
hero member
Activity: 840
Merit: 1000
B6 backside photo https://imgur.com/sIYNhVX

Ok, so, the "30C" resistor is the 20k reference one.
"
and one of the "123" resistor allows you to output 10.6v.
divided by 15 chips in a string = 0.71v at the chips
Edit:
not "123"(12k), but there should be a 1.2k resistor or close value.

1.1k resistor in place of the 1.2k will give 0.77v at the chips
1.3k resistor will give 0.65v at the chips
18k resistor in place of the 20k should also work for 0.64v at the chips
22k resistor in place of the 20k will give 0.77v at the chips

Edit, I'm aging, took me 20 minutes to figure the over/undervolt.
full member
Activity: 196
Merit: 100

i'd rather have a stable miner with slightly lower hashrate but with minimal HW error & less heat.

Exactly this. The first thing I do with all my antminers is lower the frequency slightly - it brings down the error rate, reduces power consumption making them more efficient, they run more stable & reduces wear, prolonging their usable life span. Bitmain sell their miners tuned to the limit, but I'd rather have a slightly lower hash rate but a longer lasting & more reliable/efficient miner.

Well said, Pug!

Time to sleep everyone.  Have a good day!

btw, those who just bought b8,


you got PAWNED ! price dropped heh Huh

Everything is relative.  Another way to look at it is, we got a discount from previous batch.  Also by the time others receive their Batch 9, we've made more than the cost difference in BTC.  If anything, I want a few more now than to wait another 1.5 months+ as they're not shipping out until up to Jan 30th which means it can be up until Feb when we get the Batch 9 boxes where the difficulties are also increased even further.  Time = Money......$$$

On Another subject :
Why is BitMain site keep on being down past few days?
hero member
Activity: 840
Merit: 1000
B6 backside photo https://imgur.com/sIYNhVX

Thanks.
TI's LM27402
http://www.ti.com/lit/ds/symlink/lm27402.pdf

So we may have a way to adjust voltage to the chips

I used a very similar design for string chips with LM27403.
hero member
Activity: 672
Merit: 500
legendary
Activity: 1500
Merit: 1002
Mine Mine Mine

i'd rather have a stable miner with slightly lower hashrate but with minimal HW error & less heat.

Exactly this. The first thing I do with all my antminers is lower the frequency slightly - it brings down the error rate, reduces power consumption making them more efficient, they run more stable & reduces wear, prolonging their usable life span. Bitmain sell their miners tuned to the limit, but I'd rather have a slightly lower hash rate but a longer lasting & more reliable/efficient miner.

Well said, Pug!

Time to sleep everyone.  Have a good day!

btw, those who just bought b8,


you got PAWNED ! price dropped heh Huh
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader

i'd rather have a stable miner with slightly lower hashrate but with minimal HW error & less heat.

Exactly this. The first thing I do with all my antminers is lower the frequency slightly - it brings down the error rate, reduces power consumption making them more efficient, they run more stable & reduces wear, prolonging their usable life span. Bitmain sell their miners tuned to the limit, but I'd rather have a slightly lower hash rate but a longer lasting & more reliable/efficient miner.

Well said, Pug!

Time to sleep everyone.  Have a good day!
sr. member
Activity: 266
Merit: 250

i'd rather have a stable miner with slightly lower hashrate but with minimal HW error & less heat.

Exactly this. The first thing I do with all my antminers is lower the frequency slightly - it brings down the error rate, reduces power consumption making them more efficient, they run more stable & reduces wear, prolonging their usable life span. Bitmain sell their miners tuned to the limit, but I'd rather have a slightly lower hash rate but a longer lasting & more reliable/efficient miner.
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
Not only that & to those who can adjust their voltage, might be worth to give it a try to up the voltages in very small steps.

just back to basic oc'ing techniques.

more processing power = more power needed (volt) but also = more heat = more errors = shorter lifespan

so it is up to individual to decide how much to add.

of course the lower the hw error, the higher the hashrate, the least heat is always the best result.

take it 1 step at a time BUT remember warranty is gone !

if no options (adjusting voltage) then lowering will definitely help.

i'd rather have a stable miner with slightly lower hashrate but with minimal HW error & less heat.

+1
legendary
Activity: 1500
Merit: 1002
Mine Mine Mine
Any of you that may have a slightly under performing S7 Batch 8 may consider the following:

I noticed I had to go to "Miner Configuration" several times and click "save & apply" several times to get several S7's batch 8 to jump from approximately 1,700 GH/s to at least 4,600 GH/s.  It would also take a while to get to 4,600 GH/s while the HWE's slowly ticked down to reach only 4,600 GH/s.  I under clocked to 693 MHz on those that had high HWE's with 4,550 to 4,630 GH/s maximum hash. 

When I under clocked to 693 MHz, my HWE's dropped from upper 0.0300's to 0.0060's [in most cases] and my hash rate went up from 4,550 - 4,630 range to 4,670 - 4,680 range.  You might have to go to  "Miner Configuration" again and click "save & apply" to get it to going above 1,700 GH/s after under clocking but in most cases you will not have to do this multiple times after under clocking.  I found that rebooting multiple times did not get the tedious rigs [With high HWE's and sub par hash rate] above 1,700 GH/s and the "Miner Configuration" with "save & apply" worked much better.  Especially after under clocking.

So, in summery, under clock one level down from 700 MHz to 693 MHz for better results with the batch 8's you have with higher HWE's and sub par hash rate.  This will make your hash rate better, your HWE's lower and your rig run a little cooler.

David
+1

not only that & to those who can adjust their voltage, might be worth to give it a try to up the voltages in very small steps.

just back to basic oc'ing techniques.

more processing power = more power needed (volt) but also = more heat = more errors = shorter lifespan

so it is up to individual to decide how much to add.

of course the lower the hw error, the higher the hashrate, the least heat is always the best result.

take it 1 step at a time BUT remember warranty is gone !

if no options (adjusting voltage) then lowering will definitely help.

i'd rather have a stable miner with slightly lower hashrate but with minimal HW error & less heat.
hero member
Activity: 770
Merit: 523
Any of you that may have a slightly under performing S7 Batch 8 may consider the following:

I noticed I had to go to "Miner Configuration" several times and click "save & apply" several times to get several S7's batch 8 to jump from approximately 1,700 GH/s to at least 4,600 GH/s.  It would also take a while to get to 4,600 GH/s while the HWE's slowly ticked down to reach only 4,600 GH/s.  I under clocked to 693 MHz on those that had high HWE's with 4,550 to 4,630 GH/s maximum hash.  

When I under clocked to 693 MHz, my HWE's dropped from upper 0.0300's to 0.0060's [in most cases] and my hash rate went up from 4,550 - 4,630 range to 4,670 - 4,680 range.  You might have to go to  "Miner Configuration" again and click "save & apply" to get it to going above 1,700 GH/s after under clocking but in most cases you will not have to do this multiple times after under clocking.  I found that rebooting multiple times did not get the tedious rigs [With high HWE's and sub par hash rate] above 1,700 GH/s and the "Miner Configuration" with "save & apply" worked much better.  Especially after under clocking.

So, in summery, under clock one level down from 700 MHz to 693 MHz for better results with the batch 8's you have with higher HWE's and sub par hash rate.  This will make your hash rate better, your HWE's lower and your rig run a little cooler.

David
+1

For me this works with 4.86 batches as well.
member
Activity: 71
Merit: 10
Anyone look at the efficiencies of the later batches?

I was talking to my buddy who has a bunch of Batch 6, 4.05TH ones running at 4.4-4.5TH but only using 1150-1190Watts at the wall with EVGA 1300G2's.
He told me a few were even clocked at 4.2TH when he received it from factory at 625MH.

So that works out to be more efficient than all the other batches, no? I'm talking about real world @ wall usage , not the BS they usually advertise.
Maybe Batch 6's are the ones to get and Batch 7-8 is to be avoided like the plague as a LOT of them are having issues.




Batch 1 uses 1210 W to get 4.86 TH/s = Actual Efficiency: 0.2490 J/GH   
Batch 2 uses 1160 W to get 4.66 TH/s = Actual Efficiency: 0.2489 J/GH
Batch 3 uses 1210 W to get 4.86 TH/s = Actual Efficiency: 0.2490 J/GH   
Batch 4 uses 1160 W to get 4.66 TH/s = Actual Efficiency: 0.2489 J/GH
Batch 5 uses 1210 W to get 4.86 TH/s = Actual Efficiency: 0.2490 J/GH   
Batch 6 uses 1042 W to get 4.05 TH/s = Actual Efficiency: 0.2572 J/GH
Batch 7 uses 1278 W to get 5.06 TH/s = Actual Efficiency: 0.2573 J/GH     
Batch 8 uses 1293 W to get 4.73 TH/s = Actual Efficiency: 0.2734 J/GH

It is all about efficiency 24 hours a day!  

Looks to me that the Efficiency is going down.

Is that actual measured watt usage or what is claimed by Bitmain.
My B1 used 1280-1290watts at full blast (Just around 4.85TH.

Hmm, that's not that good.
My B1 is at 625 with 1230-1235 watts running 5017GH/s
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
Don't forget to check the package for your Christmas gift  Wink

Yes, sir.  I saw that.  Thanks!  It's like you read my mind.




EDIT:  I know it's hard to see everything from this view.  This is only temporary until I get the power upgrade.  Nineteen (19) are hashing away at the moment.  Fifteen (15) more are sitting in this closet until the upgrade.  All of this will be set up looking organized and clean after the upgrade.  Unfortunately, I will have to take all of this down for the install in case the inspector decides he wants to see if the rigs are UL listed.  Also, I need to have it all out of the way to install more receptacles.  

I'm going to Home Depot tomorrow to look for something metal I have in mind for the IBM 2880's to slide into and have the rigs sitting on top of it instead of having it set up the present way.  I also need to adjust the shelves down and add another one for 6 shelves in total on each rack.  There will be two racks inside this mining closet.


hero member
Activity: 840
Merit: 1000
Anyone running B8 higher than 700M Frequency to test results?

yes,

I still need to raise voltage, it's currently running off a bladecenter H 2880w PSU, will switch to a 4k later.
It ran at 775 all night long without any issue

NICE...

Thanks for sharing, J4bberwock.  I'll need to take some of the IBM 2000 BB setups from you in a couple of months.  It will be another 3 to 4 weeks before I have my power upgraded to justify buying more PSU's.  I decided to go with what you engineered and designed for the 2000 BB once the 2880's I have are used up.  I have 18 of them.  I thought I had 20 but I don't.

By the way,

Thanks for the prompt shipment!  I received it today.  GREAT communication and service, Sir!

Cheers,

David

EDIT:  I'll order those PSU's the moment I get a work permit to upgrade power.  Looks like I'm going with 800 amp service.  1000 amp will be too expensive.

Don't forget to check the package for your Christmas gift  Wink
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
Anyone running B8 higher than 700M Frequency to test results?

yes,

I still need to raise voltage, it's currently running off a bladecenter H 2880w PSU, will switch to a 4k later.
It ran at 775 all night long without any issue

NICE...

Thanks for sharing, J4bberwock.  I'll need to take some of the IBM 2000 BB setups from you in a couple of months.  It will be another 3 to 4 weeks before I have my power upgraded to justify buying more PSU's.  I decided to go with what you engineered and designed for the 2000 BB once the 2880's I have are used up.  I have 18 of them.  I thought I had 20 but I don't.

By the way,

Thanks for the prompt shipment!  I received it today.  GREAT communication and service, Sir!

Cheers,

David

EDIT:  I'll order those PSU's the moment I get a work permit to upgrade power.  Looks like I'm going with 800 amp service.  1000 amp will be too expensive for my present budget.  1,000 amp require too many addition add on's with National Electrical Code to put it out of budget.
hero member
Activity: 840
Merit: 1000
Anyone running B8 higher than 700M Frequency to test results?

yes,



I still need to raise voltage, it's currently running off a bladecenter H 2880w PSU, will switch to a 4k later.
It ran at 775 all night long without any issue
legendary
Activity: 1302
Merit: 1318
Technical Analyst/Trader
Any of you that may have a slightly under performing S7 Batch 8 may consider the following:

I noticed I had to go to "Miner Configuration" several times and click "save & apply" several times to get several S7's batch 8 to jump from approximately 1,700 GH/s to at least 4,600 GH/s.  It would also take a while to get to 4,600 GH/s while the HWE's slowly ticked down to reach only 4,600 GH/s.  I under clocked to 693 MHz on those that had high HWE's with 4,550 to 4,630 GH/s maximum hash.  

When I under clocked to 693 MHz, my HWE's dropped from upper 0.0300's to 0.0060's [in most cases] and my hash rate went up from 4,550 - 4,630 range to 4,670 - 4,680 range.  You might have to go to  "Miner Configuration" again and click "save & apply" to get it to going above 1,700 GH/s after under clocking but in most cases you will not have to do this multiple times after under clocking.  I found that rebooting multiple times did not get the tedious rigs [With high HWE's and sub par hash rate] above 1,700 GH/s and the "Miner Configuration" with "save & apply" worked much better.  Especially after under clocking.

So, in summery, under clock one level down from 700 MHz to 693 MHz for better results with the batch 8's you have with higher HWE's and sub par hash rate.  This will make your hash rate better, your HWE's lower and your rig run a little cooler.

David
legendary
Activity: 1120
Merit: 1001
"This particular item is a collection of computer processors called ASICs that perform a specific computational task. Alone, this component does nothing. Once I assemble other pieces and attach the ASICs it solves complex computer math problems. In the most simplest of terms, it is just a computer component."

Saw someone else stated this. Should I try this ?

Yes, it worked for them.  You may want to put the following in parentheses after "ASIC": (Application Specific Integrated Circuit) with everything else you quoted.

Keep us updated please...  It sounds too much like a damn police State to me.  The whole damn world is slowly turning into a damn Police State by all these damn progressive leaders.

I'm screwed.
My antminer are still held by customs Sad
I mailed  BITMAIN to give me 2 invoices instead of 1, each of them on different names
This way I would have to import only 2 pieces on my name and the other 2 pieces on my father's name
Do you think this would work ??
legendary
Activity: 2030
Merit: 1076
BTCLife.global participant
Here's a quick binary for the S5 based on bitmain's existing code which will ignore any queue parameter, not discard stales, should be able to ramp up smoothly if you find yourself on a very low diff pool, and use a little less CPU:

http://ck.kolivas.org/apps/cgminer/antminer/s5/4.9.0-150105/cgminer

Binaries will only be temporary so will not survive a machine reboot.

The following will change the cgminer binary for you (set the appropriate IP address), the default root password is "admin":

Code:
ssh 192.168.1.x -l root
cd /tmp
wget http://ck.kolivas.org/apps/cgminer/antminer/s5/4.9.0-150105/cgminer
chmod +x cgminer
mv /usr/bin/cgminer /usr/bin/cgminer.bak
cp cgminer /usr/bin
/etc/init.d/cgminer.sh restart

There should be a more comprehensive merge in the future into mainline cgminer, hopefully by Kano. Bitmaintech has provided us both with S5s to support cgminer development.

Can we just take it and use it on S7s "as is"?
Jump to: