What happened to the Antminers with the 5.05 hash rate?

[ChrisHaefner]

Hi,
Does anyone know why Bitmain made only one batch of Antminer S7's with the 5.05 TH rate? Was this batch 5? I'd love to hear some experiences and reviews of these machines. Also if anyone has some leads on new to lightly used units of this batch I would be interested.  Thank you!

Posted From bitcointalk.org Android App

[VirosaGITS]

Hi,
Does anyone know why Bitmain made only one batch of Antminer S7's with the 5.05 TH rate? Was this batch 5? I'd love to hear some experiences and reviews of these machines. Also if anyone has some leads on new to lightly used units of this batch I would be interested.  Thank you!

Posted From bitcointalk.org Android App

It was Batch 7;



They were "experimenting" later on in the batches they now build S7's with less chips that are overvolted and overclocked to save money. At the cost of less efficiency.

[RichBC]

Hi,
Does anyone know why Bitmain made only one batch of Antminer S7's with the 5.05 TH rate? Was this batch 5? I'd love to hear some experiences and reviews of these machines. Also if anyone has some leads on new to lightly used units of this batch I would be interested.  Thank you!

Posted From bitcointalk.org Android App

It was Batch 7;



They were "experimenting" later on in the batches they now build S7's with less chips that are overvolted and overclocked to save money. At the cost of less efficiency.

Also worth adding that 5.05TH / Batch 7 was the last of the 162 Chip Systems. Batches 6,8,9 and probably all future batches are 135 Chip Systems with a Buck Controller fixing the Core voltage.

Rich

[Laviathon]

I'm getting ready to order a couple B9's I guess.  Just little concerned with all the people having the problems out of batch 8.  I did not see all the problem posts with the batch 5 stuff.  Wonder If I should look for some of those over the new ones. 

[sidehack]

Yeah, awful nice of them to start making their products worse wasn't it?

[philipma1957]

My batch 2 freq rated at freq 575 is outstanding.

It can do as high as freq 631 does 5120 gh at .26 watts per gh.

On the other end if you run it with a titanium evga t2 at freq 575 you get 0 hardware errors and .246 watts per gh.

[VirosaGITS]

I'm getting ready to order a couple B9's I guess.  Just little concerned with all the people having the problems out of batch 8.  I did not see all the problem posts with the batch 5 stuff.  Wonder If I should look for some of those over the new ones. 

I don't know, my B8 is fine so far,  not sure if B8 has just been bigger, thus more vocal people minority reported problems. Regardless, its clearly "made in China stuff". Its cheap and cost effective, just not fully reliable.

Yeah, awful nice of them to start making their products worse wasn't it?

The price sure went down, too. But this kind of thing would of been easier to swallow, imo, if there was volt control, as to be able to make the most of what they're giving us, after their stunt. Right now the way its presented, many don't realize the reason why the later batch has more or less the same hashrate as the previous batch.

Now their "Power Efficiency: 0.25 J/GH" is plain false.

[sidehack]

Oh yes, quite.

[ChrisHaefner]

Thank you!!! Finally a solid answer and an explanation of the price change.  I appreciate it.

Posted From bitcointalk.org Android App

[Laviathon]

My batch 2 freq rated at freq 575 is outstanding.

It can do as high as freq 631 does 5120 gh at .26 watts per gh.

On the other end if you run it with a titanium evga t2 at freq 575 you get 0 hardware errors and .246 watts per gh.

Wonder what the suposid gain was for making them at 700 out of the box.   What are new ones looking at power wise if not .25.    I know originally people were running them with evga 1300 watts.  Now that won't handle it if I understand correctly.

[dogie]

Yeah, awful nice of them to start making their products worse wasn't it?

Its preferable to the alternative of SP20-esque "its 5TH +/- 10%, honest".

[philipma1957]

Yeah, awful nice of them to start making their products worse wasn't it?

Its preferable to the alternative of SP20-esque "its 5TH +/- 10%, honest".

Well  no one really got 1700

And at 1600 it burned power like a mofo.

But at 1300 it was good to go.

Frankly both the sp20 and the s-7 are quilty of false advertisement .

But some s-7s are better then advertised my batch 2 is a perfect example of better then it's specs.

Not sure if any sp20s ran at 1860gh which would be +10%

But my s-7 does 5100 while specs say 4660 which is  + 10%

[Swimmer63]

Yeah, awful nice of them to start making their products worse wasn't it?

Its preferable to the alternative of SP20-esque "its 5TH +/- 10%, honest".

Well  no one really got 1700

And at 1600 it burned power like a mofo.

But at 1300 it was good to go.

Frankly both the sp20 and the s-7 are quilty of false advertisement .

But some s-7s are better then advertised my batch 2 is a perfect example of better then it's specs.

Not sure if any sp20s ran at 1860gh which would be +10%

But my s-7 does 5100 while specs say 4660 which is  + 10%

Agreed.  The only miner I ever bought that lived up to its speed was from the old ASIC MINER.

[sidehack]

To my knowledge, they didn't really change the later batches of SP20 to run worse than the early ones. That's what I meant about making things worse - not worse than advertised, but actually changing the design so the same "product" is actually a different machine with worse specs. Saves 'em money, I guess, and reduces the effective lifetime of the machine (which in turn makes them more money because they get replaced that much faster) so it's a win/win for Bitmain and everyone else just kinda has to deal with it.

[VirosaGITS]

My batch 2 freq rated at freq 575 is outstanding.

It can do as high as freq 631 does 5120 gh at .26 watts per gh.

On the other end if you run it with a titanium evga t2 at freq 575 you get 0 hardware errors and .246 watts per gh.

Wonder what the suposid gain was for making them at 700 out of the box.   What are new ones looking at power wise if not .25.    I know originally people were running them with evga 1300 watts.  Now that won't handle it if I understand correctly.

The gain was to use less chips to attain the same hash or so. So for the Chinese's "Full Volt Full speed" portfolio, cost effective BUILD, they did well, since they take 135 chips now to reach 4.73TH/s instead of 162 or whatever.

So their costs went down and we could argue that the price we pay went down too. But the end result is .29~ efficiency instead of .25, since the chips need to run at a higher voltage. (And very well maybe a higher failure rate, but i can only speculate.)

And the EVGA 1300 watts can do it, albeit at reduced efficiency. I estimate about up to 1480w at the wall.

My EVGA 1600 P2 does 1415 watts which equate to 1295w DC, but i do not have the fan running at full speed.

[NotFuzzyWarm]

My batch 2 freq rated at freq 575 is outstanding.

It can do as high as freq 631 does 5120 gh at .26 watts per gh.

On the other end if you run it with a titanium evga t2 at freq 575 you get 0 hardware errors and .246 watts per gh.

Wonder what the suposid gain was for making them at 700 out of the box.   What are new ones looking at power wise if not .25.    I know originally people were running them with evga 1300 watts.  Now that won't handle it if I understand correctly.

The gain was to use less chips to attain the same hash or so. So for the Chinese's "Full Volt Full speed" portfolio, cost effective BUILD, they did well, since they take 135 chips now to reach 4.73TH/s instead of 162 or whatever.

So their costs went down and we could argue that the price we pay went down too. But the end result is .29~ efficiency instead of .25, since the chips need to run at a higher voltage. (And very well maybe a higher failure rate, but i can only speculate.)

And the EVGA 1300 watts can do it, albeit at reduced efficiency. I estimate about up to 1480w at the wall.

My EVGA 1600 P2 does 1415 watts which equate to 1295w DC, but i do not have the fan running at full speed.

Running with the Bitmain 1600w PSU my s7b8's, as reported by my UPS's are all pulling 1,220-1,255w from the AC plug. Running stock clock speed all have been pretty rock solid at a daily average of 4.7-4.9TH/s sometimes hitting as high as 5.8THs (and occasional lows of 3.8TH/s) over a 10min period as reported by my pool.

[VirosaGITS]

Running with the Bitmain 1600w PSU my s7b8's, as reported by my UPS's are all pulling 1,220-1,255w from the AC plug. Running stock clock speed all have been pretty rock solid at a daily average of 4.7-4.9TH/s sometimes hitting as high as 5.8THs (and occasional lows of 3.8TH/s) over a 10min period as reported by my pool.

That is strange, even at 100% efficiency, there numbers do not add up. The DC draw would be higher than the AC draw you are reporting. It would be cool if the B8's were that efficient, but they certainty are not here for me.

[notlist3d]

My batch 2 freq rated at freq 575 is outstanding.

It can do as high as freq 631 does 5120 gh at .26 watts per gh.

On the other end if you run it with a titanium evga t2 at freq 575 you get 0 hardware errors and .246 watts per gh.

Wonder what the suposid gain was for making them at 700 out of the box.   What are new ones looking at power wise if not .25.    I know originally people were running them with evga 1300 watts.  Now that won't handle it if I understand correctly.

The gain was to use less chips to attain the same hash or so. So for the Chinese's "Full Volt Full speed" portfolio, cost effective BUILD, they did well, since they take 135 chips now to reach 4.73TH/s instead of 162 or whatever.

So their costs went down and we could argue that the price we pay went down too. But the end result is .29~ efficiency instead of .25, since the chips need to run at a higher voltage. (And very well maybe a higher failure rate, but i can only speculate.)

And the EVGA 1300 watts can do it, albeit at reduced efficiency. I estimate about up to 1480w at the wall.

My EVGA 1600 P2 does 1415 watts which equate to 1295w DC, but i do not have the fan running at full speed.

Running with the Bitmain 1600w PSU my s7b8's, as reported by my UPS's are all pulling 1,220-1,255w from the AC plug. Running stock clock speed all have been pretty rock solid at a daily average of 4.7-4.9TH/s sometimes hitting as high as 5.8THs (and occasional lows of 3.8TH/s) over a 10min period as reported by my pool.

My B8 stay's at 4.7 almost always.  It seems they have done a good job with latest design of getting a product that is pretty consistent on speed.  It is cranked up on freq, so OC is pretty much out of question (at least on how far I would push it).

The ones with 5T and or ones with more chips, I don't think we will see again.  They are able to use less chips now and this lowers costs, makes profits higher.   And hopefully that means they can sell for lower.    But I think 4.7T version is here to stay.