Antminer Batches?

[paulb787]

I have an antminer s7 and unfortunately like it so much I think I will be buying another lol. I have a batch 3 and I have seen a batch 1 for sale. Any difference between the batches? They are both 4.86 is that all that matters?

Thanks so much,
Paul

[1715564237]

1715564237

[1715564237]

1715564237

[philipma1957]

I have an antminer s7 and unfortunately like it so much I think I will be buying another lol. I have a batch 3 and I have seen a batch 1 for sale. Any difference between the batches? They are both 4.86 is that all that matters?

Thanks so much,
Paul

my batch one is not 4.86th

on a good day it gets 4.70th

[carlosmnk]

Batch 1: It must do 4.86 th/s, but there are "some" (a lot maybe) people reporting 4.66 th/s miners
Batch 2: It must do 4.66 th/s
Batch 3: It must do 4.86 th/s
Batch 4: It must do 4.66 th/s
Batch 5: It must do 4.86 th/s
Batch 6: IT says it do 4.05 th/s and with same eficience (135 chips instead 162 chips version), but some here think it will have worse eficience (perphaps +10% draining power)

Also there are some miners reporting 4.45 th/s (you can but it at itopshop.net ), and bitmaintech was released a new firmware with 3 variants: 4.86 / 4.66 / 4.45 ....

I think this miner have too many variances, and in near future, when we will see an S7 used we will don't know really what can do.... But it runs great...  

[MarkAz]

My experience mirrors carlosmnk, I have 2 Batch 1's, and they're both a bit of a dog doing ~4.6 w/ 0.04 HWE, but I have 7 Batch 3's that are far better, all doing 4.8 with HWE 0.002 at worse.

BTW; on the itop site, it says Batch 3, but the spec shows 4.05, so it's batch 6.  Zoomhash lists 4.86, but I had a bunch on order with them and they just cancelled them all without warning and then raised the prices on the site.  Sucks because I could have just purchased them directly from Bitmain but I thought my experience would have been better with Zoomhash.

As carlos says, it makes the secondary market on these things a minefield - who knows how the machine will perform, which batch, etc.

[QuintLeo]

Batch 6 WILL be worse efficiency, as they're running it at a higher voltage. My best guess is more like 0.3W/GH vs the 0.25 spec on the earlier batches.

[philipma1957]

Batch 6 WILL be worse efficiency, as they're running it at a higher voltage. My best guess is more like 0.3W/GH vs the 0.25 spec on the earlier batches.

not true but could be true.

my batch 2 is a stud and the boards read 575 freq.

my batch 1 sucks

I will post photos.

the batch 1 does 4760 at freq 593---errors are 0.0141%
the batch 2 does 4909 at freq 606---errors are 0.0009%

I can get better numbers if I feed hot volts :

 12.1 is going into the batch two 
11.94 is going into the batch one

the batch 6 may have the widest volt range of all the batches--------------  if true it has an interesting value.

a psu like this one

https://www.trcelectronics.com/ecomm/pdf/rsp1500.pdf

could allow for under volt  & down clock or    over volt  & over clock

so buying used you won't have any idea of what you get. unless you see the gear running.
you need to see the gear and read the input volts to all 9 blade jacks to know what you are buying.

and  I have done extensive testing you can not cheat on the 9 blade jacks 9 separate cables. on some psu's the dc drops a lot when you split the pcie cables.  you go to 11.8 and the s-7 starts to do poorly.
So a guy can say his batch 1 sucks and gives him 4500
and the truth is his power sucks and he could do more like 4700 with better psu's
I will not buy any s-7's via resale unless I test the s-7 first.

[Finksy]

It seems to me that even in Batch 1 there were "early" batch 1's and "later" batch 1's.  There were many people who ordered in August that received all of theirs together, and then the rest of the B1's shipped right at the end of their deadline (including my 2 which confirmed on Sep 2nd), and mine have been rock solid 4.86, and even over-clocked to >5 TH's without a hitch.  Maybe some of the late B1's were actually B2 after they found some defects? And the early B1's were more like the B3's...

[paulb787]

My batch 3 I bought on ebay does says 4795 gh/s average. is that acceptable? what else should I look for?

[MarkAz]

My batch 3 I bought on ebay does says 4795 gh/s average. is that acceptable? what else should I look for?

Yeah, that's totally within spec - the big thing is hardware error rate.  The lower the number, the better - for me, I like to see them less than 0.01, but I can live with less than 0.05.  You get unicorns that have no errors, I have two that literally have 0 all the time. 

[paulb787]

where do I find the error rate?

[RichBC]

where do I find the error rate?

Look at the pictures 4 posts back and in the middle of the Status Screen you will see a line marked HW. First number is the number of HW errors and further down the line the HW error % that is being referred to.

Although how someone can drop $1800 on an S7 without any understanding of where you can see the HW error rate escapes me..... 

Rich

[Biodom]

Re batches;

my early (8/30) batch 1 is actually lower (in a shorter box), performed at spec 4.84 Th.
my batch2 (10/22) is as described with 575 sticker, but is better at 575 (4.64 Th). I upgraded this one to 575 firmware, which seem to improve it in error and caused slightly lower fan speed.

my batch2 (10/16) has no sticker, has practically zero error rate at default 575, but could run at 600 with 0.0050% error, which is OK.

This one is an enigma. For example, should I update it with 575 firmware or 600, or none. None would be OK at home, but it is going to hosting, although I can always put it on 80% manual (to prevent the internet loss burnout situation).

Phil, did you update you '575' stud to new firmware? If yes, did you choose 575 or 600 firmware?
thanks

[Biodom]

where do I find the error rate?

Look at the pictures 4 posts back and in the middle of the Status Screen you will see a line marked HW. First number is the number of HW errors and further down the line the HW error % that is being referred to.

Although how someone can drop $1800 on an S7 without any understanding of where you can see the HW error rate escapes me.....  

Rich

when i started i couldn't figure out what WU is as well.
Besides, for some $1.8k is peanuts (I am NOT one of those).
People drop $10-40k on a single dinner  
at some point someone will make a pink gold covered miner for $50K if btc will become popular.

[paulb787]

Lol. I am still learning. This forum has been a huge help. It did make me nervous to get one on ebay. like someone earlier said they could be selling it because there not happy with the performance.

But my HW is 8727 after 24HRS and 0.0092% .. is this normal? What is the error rate? thank u for all your help.

http://s17.postimg.org/4kro7v2or/Screenshot_8.jpg

http://postimg.org/image/4kro7v2or/

[dogie]

Batch 6 WILL be worse efficiency, as they're running it at a higher voltage. My best guess is more like 0.3W/GH vs the 0.25 spec on the earlier batches.

Do we actually know they're using shorter chains in this smaller one?

[carlosmnk]

Batch 6 WILL be worse efficiency, as they're running it at a higher voltage. My best guess is more like 0.3W/GH vs the 0.25 spec on the earlier batches.

Do we actually know they're using shorter chains in this smaller one?

really no, i think

[dogie]

But my HW is 8727 after 24HRS and 0.0092% .. is this normal? What is the error rate? thank u for all your help.

HW errors are results which don't make sense and are discarded. Consider them random occurrences which happen at a higher rate when things aren't performing rate. <1% is generally inconsequential.

[paulb787]

got it! so generally anything under 1% is acceptable. my 0.0093% would qualify as normal. If it read 1.0000% or more then it would be a problem. It could cause reboots or be a sign of failing chips. Some antminers will have more HW errors then others and it is just the way it was built.


So if I were to buy my second one.. I should stay away from a batch 1 because some have reported that it does not run at spec. Which is + or - 5%. Should I be upset that mine is only 4.795 TH/S and 0.0093%?

Thank you again for the help!

[dogie]

Should I be upset that mine is only 4.795 TH/S and 0.0093%?

Not really, its all random variance. If you're at an unusual pool you might get a bit more consistent hash rate at a larger pool. Might be worth a test.

[philipma1957]

Re batches;

my early (8/30) batch 1 is actually lower (in a shorter box), performed at spec 4.84 Th.
my batch2 (10/22) is as described with 575 sticker, but is better at 575 (4.64 Th). I upgraded this one to 575 firmware, which seem to improve it in error and caused slightly lower fan speed.

my batch2 (10/16) has no sticker, has practically zero error rate at default 575, but could run at 600 with 0.0050% error, which is OK.

This one is an enigma. For example, should I update it with 575 firmware or 600, or none. None would be OK at home, but it is going to hosting, although I can always put it on 80% manual (to prevent the internet loss burnout situation).

Phil, did you update you '575' stud to new firmware? If yes, did you choose 575 or 600 firmware?
thanks

I left it alone.

it works so well:

 freq 606 
errors are now under 0.0001%
hash rate is 4910.78
uses 1200 watts at the pdu meter   but the meter reads amps so 5.0 amp reading could be as high as 5.09  before it goes to 5.1.



so my watts are 1200 to 1221.6     this means at the pdu I am getting .244 to .248 a gh
I can't expect any  better so I left it alone.