ANTMINER S7 is available at bitmaintech.com with 4.86TH/s, 0.25J/GH

[notlist3d]

you think those temps are okay?

I don't but maybe I am wrong.

Batch 8 with a slight OC.  Anything higher raises the errors above .01

I would want to know ambient temp.  But that is getting up there.  What does it run not OC'ed??

I to think it is most likely to high.  I don't think Batch 8 was made to OC as much as some might want it to, it simply is already very high on frequency.  And it's sure not worth losing warranty for a slight OC.

The room this is in has an AC unit which blows outside air all day which is cold during the day and very cold during the night.  The temps in that room get up to about 73-75°F in the afternoon which is about when that screen shot was taken.  Here it is now.  Fans are at 60%.  It is only a couple hours in the afternoon when the hashing board hits the low 70's.  It will be below 70 in another hour or so.



Here is a batch 5 with fans at 60% sitting next to it.



I could bring the temps down by raising the fan speed.  There is also an oscillating fan in the room and I have found that by pointing it at the shelf with the miner intakes I can lower the temps also but I didn't really think a few hours in the low 70's was a big deal.

Likely a few hours you are right did not do any damage.  It would need to be more long term I think for it to take effect.

But with bitmains current policy if you are caught you lose warranty over a little bit of speed increase.  That is part I don't think it is worth even though chances of being caught are slim..

[1715277886]

1715277886

[1715277886]

1715277886

[1715277886]

1715277886

[klondike_bar]

^honestly, the chips can handle >70C without much issue, I imagine the 80C shutdown is more for incidents (such as the fan failing or getting obstructed), but that if you used it properly they could likely run stable at up to 90C for days or weeks without failing.

The dangers come from rapid thermal changes though, such as a chip turning on/off, or the fan ramping up and down. thats when the solder and glue fails most easily.

[nhando]

Hehe, my Avalon 6 loves to run in the 72-76 range even when the garage is very cool and the S7 in the same ambient temperature would be running in the 58-62.  =)  

[elokk]

is https://bitcoinwisdom.com/bitcoin/calculator a good place to calculate ROI? i keep messing with this thing and still cant find ROI.....everytime i do it im in the hole $700+ after 200+ days. these things need to be cheaper. and with the reliability issues of late, i think im better off just buying a bunch of btc and seeing if it increases in value over those 200 days...when stuff lasts 2 weeks and burns up or fails..... that means it is total crap!

I'm curious...  Does your electricity rate per kWh need to be cheaper?

5.9 cents a KW. which is low for most of the USA....

And you can't see ROI?

Can you not see future price of bitcoin going up considerably just before the block halving?

Of course, I have plans on moving soon to an area with $0.0285 to $0.0325 per kWh.  That's the main reason I'm loading up on rigs.  However, I'm not moving until I have 200+ S7's.  It could be end of July [After block halving] or it could be the last quarter of 2016.  It depends on how certain things play out as to the timing of my move.

Where are you looking to move that has those low electric rates?

[klondike_bar]

you think those temps are okay?

I don't but maybe I am wrong.

Batch 8 with a slight OC.  Anything higher raises the errors above .01

Here are all of the measurements put together:

Calculation would be TH/s per MHz:

4.933TH / 731MHz = 6.748290013679891
4.867TH / 725MHz = 6.713103448275862
4.730TH / 700MHz = 6.757142857142857

RESULTS:  You get more TH/s at 700MHz.   The only other consideration would be the Wattage per TH/s.  Very good HW, but be cautious on the temps.

hashrate and frequency are linearly related, with a fair share of variance too.
technically your above image shows a better average hashrate (the 5s hashrate can vary greatly)

4.911TH / 731MHz = 6.718  (based on 30hrs)
4.867TH / 725MHz = 6.713 (based on 40hrs)

[RichBC]

Here are all of the measurements put together:

Calculation would be TH/s per MHz:

4.933TH / 731MHz = 6.748290013679891
4.867TH / 725MHz = 6.713103448275862
4.730TH / 700MHz = 6.757142857142857

RESULTS:  You get more TH/s at 700MHz.   The only other consideration would be the Wattage per TH/s.  Very good HW, but be cautious on the temps.

hashrate and frequency are linearly related, with a fair share of variance too.
technically your above image shows a better average hashrate (the 5s hashrate can vary greatly)

4.911TH / 731MHz = 6.718  (based on 30hrs)
4.867TH / 725MHz = 6.713 (based on 40hrs)

The theoretical multiplier for a 135 chip S7 is 6.75, so that should give the max possible hashrate for a given frequency.

750MHz x 6.75 = 5.053TH/s
725MHz x 6.75 = 4.894TH/s
700MHx x 6.75 = 4.725TH/s


Rich

[AriesIV10]

Here are all of the measurements put together:

Calculation would be TH/s per MHz:

4.933TH / 731MHz = 6.748290013679891
4.867TH / 725MHz = 6.713103448275862
4.730TH / 700MHz = 6.757142857142857

RESULTS:  You get more TH/s at 700MHz.   The only other consideration would be the Wattage per TH/s.  Very good HW, but be cautious on the temps.

hashrate and frequency are linearly related, with a fair share of variance too.
technically your above image shows a better average hashrate (the 5s hashrate can vary greatly)

4.911TH / 731MHz = 6.718  (based on 30hrs)
4.867TH / 725MHz = 6.713 (based on 40hrs)

The theoretical multiplier for a 135 chip S7 is 6.75, so that should give the max possible hashrate for a given frequency.

750MHz x 6.75 = 5.053TH/s
725MHz x 6.75 = 4.894TH/s
700MHx x 6.75 = 4.725TH/s


Rich

Thanks for the corrections on this!  The BIG overlying piece of information is missing...wattage

[nhando]

Yes, wattage info would be good.  What is the highest anyone OC to while able to keep it stable?  IF it's that Linear, it sounds like there's a ton of OC potential, just not sure if the wattage increase is also as linear.

[RichBC]

Yes, wattage info would be good.  What is the highest anyone OC to while able to keep it stable?  IF it's that Linear, it sounds like there's a ton of OC potential, just not sure if the wattage increase is also as linear.

Overclock potential on the 135 chip systems, Batches 6,8,9 will be limited by the available Core voltage which measured I think on a batch 6 is 10.12V which / 15 = 0.675V.

That is bang in the middle of the Bm1385 Data sheet for 600MHz, so going much higher than that on Batch 6 will need the voltage to be adjusted. I am assuming that the output of the Buck Converter on the 700MHz systems, Batches 8&9, is set higher, but have not seen a measurement yet? Something around 10.2V giving a core voltage of 0.68V would seem about right?

Adjusting the Core voltage on the 135 chip systems is a work in progress that I unfortunately cannot contribute to as i do not have one, but I am sure a way will be found soon. 


Rich

[not.you]

you think those temps are okay?

I don't but maybe I am wrong.

Batch 8 with a slight OC.  Anything higher raises the errors above .01

Here are all of the measurements put together:

Calculation would be TH/s per MHz:

4.933TH / 731MHz = 6.748290013679891
4.867TH / 725MHz = 6.713103448275862
4.730TH / 700MHz = 6.757142857142857

RESULTS:  You get more TH/s at 700MHz.   The only other consideration would be the Wattage per TH/s.  Very good HW, but be cautious on the temps.

hashrate and frequency are linearly related, with a fair share of variance too.
technically your above image shows a better average hashrate (the 5s hashrate can vary greatly)

4.911TH / 731MHz = 6.718  (based on 30hrs)
4.867TH / 725MHz = 6.713 (based on 40hrs)

Well actually the numbers there are from AriesIV10 but the screen shot is mine.  Also, the "GH/S(avg)" number is just calculated from the "WU" as far as I can tell on all antminer models.  You can tell it isn't an actual average because sometimes you can hit refresh on a regular basis and watch the "GH/S(5s)" number sitting on one side of the "GH/S(avg)" number while the "GH/S(avg)" number moves in the opposite direction.

[VirosaGITS]

So i found out that laying the S7 on its side get better even temps. For instance on the side i get 56 57 56. While standing i get 55 55 59. Anyone know if there is reason the S7 should not be laid on its side?

I am assuming that the output of the Buck Converter on the 700MHz systems, Batches 8&9, is set higher, but have not seen a measurement yet.

Rich

I got my B8 but i did not want to remove the warranty seal. I'm not sure there is a way to even open it up and inspect the miner for lose heatsink. It said i HAVE to inspect the boards prior to using it, as if the bord burn down because of it, it will not be under warranty.

So how am i supposed to open it up? That would make it pretty easy to get a couples of quick measurements.

[RichBC]

So how am i supposed to open it up? That would make it pretty easy to get a couples of quick measurements.

Don't know the answer to this, but I have read that you can remove the fan without breaking any seals? Should be possible to measure the voltage with the fan remove?

Rich

[AriesIV10]

So how am i supposed to open it up? That would make it pretty easy to get a couples of quick measurements.

Don't know the answer to this, but I have read that you can remove the fan without breaking any seals? Should be possible to measure the voltage with the fan remove?

Rich

Think about this:  If something goes wrong with this unit, they (Bitmain) will have you take the fan and back plate off of the unit.  If it comes down to having to send the hashcards back to them (Bitmain), they have you remove the hashcards and send the cards back to them.  They do not allow you to send the housing due to shipping cost.  Just think about this.

[RichBC]

Think about this:  If something goes wrong with this unit, they (Bitmain) will have you take the fan and back plate off of the unit.  If it comes down to having to send the hashcards back to them (Bitmain), they have you remove the hashcards and send the cards back to them.  They do not allow you to send the housing due to shipping cost.  Just think about this.

OK we are thinking......

[AriesIV10]

Think about this:  If something goes wrong with this unit, they (Bitmain) will have you take the fan and back plate off of the unit.  If it comes down to having to send the hashcards back to them (Bitmain), they have you remove the hashcards and send the cards back to them.  They do not allow you to send the housing due to shipping cost.  Just think about this.

OK we are thinking......

hahaha, you crack me up!   

[torepia]

So, could you guys post ur hashrate and temperatures?

I'm pretty sure you get lower hashrates if your temps are below 45-50C.
At the moment, some of my S7's are getting cooled by direct -17C outdoor air, which keeps the temp on the boards at 33C (Thats what the sensor says anyway.., not sure if this is accurate when getting blasted by -17C air). The hashrate avg is 4.7TH/s @ 625Mhz
Some other S7's are "cooled" with indoor air, which keeps the temp on the boards at 65-70C, these are hashing at 4.99TH/s avg. @ 625Mhz.

The ones cooled by outdoor air used to hash at 4.99TH/s as well.

Conclusion:
Board temps below 45-50C = Bad? ^^

[Biodom]

So, could you guys post ur hashrate and temperatures?

I'm pretty sure you get lower hashrates if your temps are below 45-50C.
At the moment, some of my S7's are getting cooled by direct -17C outdoor air, which keeps the temp on the boards at 33C (Thats what the sensor says anyway.., not sure if this is accurate when getting blasted by -17C air). The hashrate avg is 4.7TH/s @ 625Mhz
Some other S7's are "cooled" with indoor air, which keeps the temp on the boards at 65-70C, these are hashing at 4.99TH/s avg. @ 625Mhz.

The ones cooled by outdoor air used to hash at 4.99TH/s as well.

Conclusion:
Board temps below 45-50C = Bad? ^^

it's batch-dependent, or so it seems.

[not.you]

So, could you guys post ur hashrate and temperatures?

I'm pretty sure you get lower hashrates if your temps are below 45-50C.
At the moment, some of my S7's are getting cooled by direct -17C outdoor air, which keeps the temp on the boards at 33C (Thats what the sensor says anyway.., not sure if this is accurate when getting blasted by -17C air). The hashrate avg is 4.7TH/s @ 625Mhz
Some other S7's are "cooled" with indoor air, which keeps the temp on the boards at 65-70C, these are hashing at 4.99TH/s avg. @ 625Mhz.

The ones cooled by outdoor air used to hash at 4.99TH/s as well.

Conclusion:
Board temps below 45-50C = Bad? ^^

I have found it to be generally true that all of my antminers (going back to at least S3 is when I first noticed it) get better hashrates at higher temps (to a point).  And what I mean by "higher" is also model specific.  I remember setting some of the S3's so that the hot air coming out of the PSU was directed at the intake on the miner because it improved the hashrate slightly when they were running in a room that was  60-65 °F ambient.  I should maybe also point out that on my S3's I had removed the blue fan wire so they ran at 100% fan speed all the time as well.  I have an S5 located in a basement crawlspace (that has vent opening to the outside) that gets lower hashrates now that it is winter and the ambient temp is lower where it is. I am not sure what the optimal temps on the S7's are yet but looks to me so far like in the 60's as reported by the GUI is where I am having best results.

[marvykkio]

I had to get my 3X S7 today,
I hand them on Monday,
I am angry



By the way what happens to the site bitmain? the writing is half Chinese and half English

[left-ca]

How can i turn the beeping off when the internet cuts out. Its very annoying.....my computer reconnects it automatically within a minute or two