Topic: ANTMINER S5: 1155GH(+OverClock Potential), In Stock $0.319/GH & 0.51W/GH - page 40. (Read 451011 times)

The power into the board didn't change, it still accepts in 12V. String design was achieved by removing the DC-DC and chaining the chips together - that has nothing to do with the PCI-E connections. The S3 to the S5 board retained the same dimensions (-~1cm off the top), the same mounting holes and the same 2 PCI-E connectors. It is NOT a declaration that you must use 2 PCI-E connections.


So you were running 620W over 8 PCI-E connectors (77.5W) and it melted. Maybe you should both consider that maybe its your Seasonic PSUs that could have a problem, not the rest of the world or the other 10,000s of S3s.

I noticed they are listing 25C as the ambient temperature for measuring watts. There's no mention of operating temperature range.  Is it still up to 35C?  And what kind of factors are we looking at with the higher temps.  With summer here in the US, seems like it could be a serious issue since no one will want to spend the money on a/c.

Also, they have a 33% increase in $/gh ($.31 to $.41), 13% reduction in w/gh (.51 to .44).  I guess they can charge the premiums to China because of the cheaper elec.  Not seeing a great deal for the US though.

Still an interesting miner for me.

I may accommodate 4kw miner

Curious how will compare with 2*sp3x underclocked...

Why s5+??? this should have been the s6 or s5x3 or something like that

Given a shipping cost of $90, and NO increase in difficulty, a handy table to see how ugly

A very nice idea indeed.

For the home miner it would be perfect if they were to fully populated the S5+ Hash boards moving from 16 Trios 48 chips to 18 Trios 54 chips. This would drop the Core voltage from 0.75V to 0.67V Total Hash would drop from 2574GH to 1944GH and the Power would drop from 1192W to 758W nicely in the zone for a 1200W PSU. Efficiency would improve from 0.46J/GH to 0.39J/GH. All these figure approx calculations.

I want just the center unit and the controller for 900 shipped.

I want to buy add on left or right sides for 750 a piece.


come on bitmain sell them like this!!!!!!!


solves a shit ton of issues.


and if I am right the center would do 2550 gh at 1134 watts  the lower cost would allow it to come into the usa with zero repeat zero duty issues.

I would be willing to do a long series of tests for you if you sell me a center piece with the controller. At a decent price.

I guess 900 is low but you would get a long write up showing how to run it .


What a nice way for home miners to get back into mining sell us the center piece with an  add-on option for later!!!


This pricing avoids all usa duty issues since the center piece would be well under 2500 usd price.

$3061.04 usd is what i got for a price yuan to usd for those too lazy to do this. Probably shipping is extra

man it is a beast that is for sure ...not recommended for home use indeed Indeed!

on tradeblock with my 13c kwh elec ..assuming I won the lottery and got one as a NO ROI toy I get this again from www.tradeblock.com page

7722gh

3436 watts

add your guess and shipping on the above calculator....but at 13c kwh for me at current btc price it is pretty dang ugly to say the least


oh well a good asic porn pic

I also swapped PSU from S2 because when I measured consumption from wall it was not plesent to see. Sure device worked for a while but when I put 1200W PSU problems with the device disapeared.

You probably were running those S3's in a cool environment .... I've had a few scrapes with singly powered S3 boards during the winter. The cables may be in spec when running at those speeds, but bitmain's testing is also based on average ambients, so low(er) ambients will result in better efficiency and higher power draws ....

S5+
http://shop.bitmain.cn/goods.php?id=47

bitmain should have been running 8 pin pci-e plugs since the s3..

this is from a seasonic 850 watt psu, running 2 s3s.. running at 225 speed.
this psu came with 4 PCI-E cables with 2 plugs each.. so all 4 of the pci-e plugs on the s3s were plugged in.



as you can see, it melted..

the s5 uses more power from the pci-e then the s3 does..
any plug in style the same as this will eventually melt..

https://bitcointalksearch.org/topic/m.12091187

So its confirmed.. but gotta say.. thats one giant beast of a miner. Bitmain seem to be moving more towards industrial miners with the s4+ and s5+.

Your the one who is "arbitrarily deciding"
You have someone somewhere telling you what to say is OK.
Keep being a good little doggie and doing as you are told.

Ignore specifications, they're only for he faint of heart who think failures and fires matter ...

You got a spare 1000W PSU to send me to replace the failed one that Bitmain didn't pay for on my S2?

 I didn't know that but it would explain the MANY instances I have seen mentioned of Nepture power connectors MELTING. I also find it VERY HARD to believe they got away with that on a STOCK Neptune at all. Air cooling on the connector as mentioned by someone else might help some, though.

 No, 300 watts per PCI-E isn't insane - but running a connector outside of spec increases the chance of failure, and it's NOT a linear function. More like exponential, kicks up VERY fast once you exceed the rated capasity of a connection. I'm amazed that ANY Neptune ever worked for more than a minute, if they were stressing the connections at 400 watts by default.
 For the record, I'm not fond of the idea of pushing 288 watts through a PCI-E 6-pin connector - that's too close to the edge, leaves NO margin for folks that want to overclock or folks like me that don't have A/C and do have hot summers.

 Yes, you can get away with it in a cool environment - part of the rating is an allowance for ambient temperature - but that does NOT make it a good idea.


 You say that like the board design was unchanged between the two.
 S5 is a stringed power design, S3 had voltage conversion, which implies AT LEAST a major redesign if not a "from scratch" design of the S5 hash boards specifically including the power circuitry. There was no "inherit" based on looking at the COMPLETE REDESIGN of the power setup for the S5.



 Silicon semiconductors draw MORE power as they get hotter - this is the cause of "thermal runaway". Plus the higher temperature makes the wiring run hotter even at the SAME temperature, pushing the connector closer to it's meltdown point. The "HW error reset" only affects the chips for a very short period IF at all.
 

Don't give the Chinese any ideas.

Bitmain; "Yes, as all PCI-e Cables have been running over spec, they all vaporised and burned, therefore all your contracts are void, bye."

I jest, i don't mean anything by it. ;P

Praying to God ,  that Hashnest would not burn down?
I have been there today 100TH/s



Contract End
3.    If mining farms may no longer continue operating due to change in legal status, war, earthquake, fire, electricity failure, or other acts of God, Hashnest does not assume responsibility for compensation.

If you're defaulting to the "you only run 75W because the specification says so", then good for you. The rest of us will continue doing what we've been doing for years.

Incorrect.

What bitmain MIGHT know is that for the cables and setup they used, they had a certain % failure over the short time frame of testing they did.

The PCI-e cables themselves have a specification that they are expected to work within.
Some manufacturers may build cables closer to those limits than other manufacturers, but they are still within the specifications.

No idea who told you that specifications are OK to ignore, but tell them to stop spouting that sort of trash.

Heh, my 1000+W S2 that had a 1000W Enermax from Bitmain, that says 996W on the 12V rails, that I suggested back then that was pushing it too close to the limit ... guess what? It failed early in the year. I've since then used a 1200W Seasonic Platinum on it.

Read my comment above .........

How many times must you explain that,
Neptune Cubes PCI-E plug connector is air-cooled by Cube fan itself.

So you're saying that 288W is fine but 300W is stupid.... There is no such thing as a discrete bound, this is entirely about safety factors and how close do you want to get, not "1 more W and it explodes". Using the evidence we've gathered from 1000s and 1000s of past miners we know the approximate threshold and what is pushing it too far. 300W is still fine.We know that


They didn't insist on anything, it was arbitrarily inherited from the rev 3+ S3 board which had 2 PCI-E connectors per 160W board. And its not a rare thing, look back in this thread, plenty of people (including me) running S5s on single PCI-Es.


I feel like you're arbitrarily deciding what will work or not. I've actively tried these things and I've actively measured cable temps so it doesn't really make sense for you to say its all about to explode.


Neptunes are 400W per PCI-E at stock, not 300W. And even then the failure isn't guaranteed.