Topic: ANTMINER S7 is available at bitmaintech.com with 4.86TH/s, 0.25J/GH - page 200. (Read 527809 times)

I can't remember.  Isn't that the company that helped pull off the Black Arrow scam (used to be minersource), or am I thinking of someone else?

Edit:  Oh yes.  Matt Carson, aka bobsag3.  His old scammer profile is borito4.  Maybe he has decided to go legit...

I guess I'm the odd man out here, but I've been hosting an SP20 there since March with no problems whatsoever.

Bitmain replaced my faulty boards fairly quickly, within a day of receiving the boards. Make sure to send pictures to expedite the process.

thanks for the info too...sorry to hear it
a major bummer

Thx. If they don't replace it under warranty in a timely fashion, I will not be buying another S7. Up til this incident happened, I was ready to grab one soon.

Rich addressed this in an earlier post, which see:

Thanks for your input, Rich.  It's much appreciated!!!

I think you need to take the most recent specs with a "grain of salt". They say the miner has 135 chips, which yields 45 chips on each of the 3 boards. Yet, you go and count the spots on the picture, and there are very clearly 54 ASICs shown. Those are the original boards for a 162-chip miner (i.e. Batch #1 etc.).

I think the reality gets a bit jumbled up with the various "cut and paste" actions to keep the website mostly close. I expect the website guys have every bit as much difficulty trying to keep stuff straight as we do.

Batch 8 is again with a little bit less power. How comes that batch 7 was 5.06 TH/s fast ? Is it possible, that the next patch would be more than that ?

Problem is we are in uncharted territory. S3 had a small frequency cut back on the first release, then we saw the S3+. S5 there was only one default frequency throughout. Now here we are about 10 Weeks into the S7 and we have over 8 Batches seen 6 variants and 2 different chip counts.

We are given no information as to why so all we can do is try and guess what is going on. There are only the original low res pictures and because of the warranty implications noone will take the covers off. I have seen mention of at least 3 board revisions but we cannot see if they are just cosmetic or have real changes?

Best guess has to be that there is significant variance in the way the BM1385 chip performs. It appears that this variance was only discovered when volume production started and the solution to the variance has been to produce batches with different default frequencies and latterly different chip counts to bring a voltage variable into play.

Which batch is "best" is difficult to answer, and most people will not in practice have made that choice, but will have just bought what was available when they decided to pull the trigger.

My generality would be that the higher the default frequency and the lower the core voltage the better the chip / system. So for the 162 chip system that would be Batch 7 at 625MHz and for the 135 chip system that would be Batch 8 at 700MHz. Perhaps it is no surprise that these are the latest two Batches?

However the fact that the latest two batches are the "best" in itself just raises a further question. Are they the best because Bitmain have just got together enough "good" chips to do this or have changes been made to the BM1384 chip itself?

So what about choosing between batch 7 & 8?

The core voltage in Batch 7 is 0.667V and at 625MHz this is the closest we have seen so far to the Data Sheet spec of 650MHz for that core voltage. 

Batch 8 is 700MHz but has a high core voltage of 0.8V because of the reduced chip count. This voltage is off the chart in the BM1385 Data sheet which indicates that 775MHz ought to be possible with only 0.71V on the Core.

So I think that batch 7 has the best chips and batch 8 although sounding impressive actually has a lower grade, but better than it's predecessor  Batch 6. On top of that I think they have been frugal with the truth on the efficiency of the 135 chip system. Headline is still 0.25J/GH. However on paper it calculates to 0.274J/GH, factor in their + 10% on Power and that becomes 0.301J/GH.

I think in practice with that Core Voltage it may even be worse than this and will be at least be something that can be measured without taking the lid off. 

So the above just my opinion & conclusions. I have no more real information than anyone else and do not even possess an S7.... As to what is actually going on or what flavour Tomorrows batch will be then your guess is as good as mine?

Rich

Batch 8 is up?

Thanks for the info.  So, my going manual now should do it.  Sorry to hear about your rig and the delays you will have to contend with.  I hope it gets worked out for you soon.

It was on 10/23 firmware and automatic fan.

I doubt that very seriously.  It's possible [after assembly] they keep rigs that meet certain parameters.  If I owned BITMAIN, I would keep rigs that met certain parameters as well.  Wouldn't you?  Does this mean all the rest are trash?  No...

I prefer the batches with more strings that results in better power efficiency rather than the 4.0 TH rig with fewer strings and less power efficiency.  I know it says batch 6 is the same .25 watts per GH/s.  However, I have to see it to believe it.

I dont understand why they are changing it so much ? Are they keeping the good chips for them selves and then just selling off all the poorly binned chips to us ?

Why not just have made it a 4TH machine from the start and not have it so highly strung. Users can OC just like before.

Uh, if this is directed at me, I did just lose internet.  I guess it's misleading with powerline, but powerline in this context is the networking adapter I use to connect my miners from the basement to my router in my kitchen.  Sometimes it just loses network connectivity and I have to unplug/replug it back in to get it to re-sync.

As long as you aren't constantly losing internet or something else is causing the board to go from hot to cold extremes, you should be fine.  However, you should set the fan % low enough to where you are saving a couple watts, extending the fan life, and keeping the noise levels down.  No sense setting it at 100% if 40% will give you the same temperature range.

what if they are in a very cold room (for example 10C or less during winter), would then excessive cooling (by setting fixed %) be detrimental?
anybody had an experience?

I've decided not to take any chances.  I've put most of mine on 60% fan speed.  I put 3 of them on 70% because their temperature was a bit warmer than the others when set to 60%.  I would recommend everyone else do the same.  It's better to be safe than sorry.  Especially, once the 90 day warranty passes.

lower clock speed if definitely better because you have a chance of overclocking if you want to risk it.
In that sense it is much more likely that B5 could be overclocked from 600 to up to 625 than B7 from 625 upwards.
people will know for sure re batch 5 by this Fri-St, re B4-today or tomorrow.

Which batch is better I cant decide if more chips is better, or less chips but at a higher clock speed and also if batch 7 can be overclocked.