The situation is pretty extreme... KNC has been going downhill for quite a bit of time. People are comparing them to BFL even... Despite this you should remain your cool if you visit their offices. Get a rest before going, you'll lose all your reasning if you're not cool and they'll simply kick you out from there... Swedish folks that bought from them should organize a small group to go and talk to them, otherwise they won't move much or pay attention to you, only  I've been around this KNC subject for quite some time, as I've been with all of the BFL situation... I hope everyone gets out of KNC without much trouble. I'll be watching... At least so far they have not been compared to Joshua Zipkin and AMT Miners. Josh along with AMT's partners in crime Bitmine.ch have out-scammed them all. Joshua Zipkin has fled to Bulgaria to avoid the lawsuit in-progress and other pending prosecution he will face if he ever returns to the USA. |
|
|
|
Anyone else here running the S4 at ~0.75Volt or more ? Any lessons leaned that would be nice to know? also what temps are you running them at when over-clocked? Any reported failure rate of the chip? Could be nice to know. The s4 uses the same chips as the s3/3+ and C1 use. Same rules apply provided you have enough PSU to do it and keep temps reasonable. As for failed chips, I have 37 active miners from s2 up and 9 retired S1's. None have EVER had a chip (permanently) fail since I started mining in Feb 2014. Go offline from sketchy speed/voltage yes but never actually die. By and large it is temps that cause any aging and even with that is more due to them drastically changing often (leading to material fatigue) than steady-state operation. Highest I've ran are under 65c when pushed but have since dropped to more moderate oc and back to 0725 for voltage. Reason? The efficiency of hash rate vs power drops like a rock and power ain't cheap. Plus, got a nice size farm so running everything at or just above stock speeds is no biggie as far as total farm hash rate. |
|
|
|
is it possible to power an s5 with a 550W gold psu? Seasonic G-550 80+ Gold Modular Only if underclocked. Using HP server supplies my S5's running stock speed pull 605-610 watts from the wall as reported by the UPS's on some of them. One NEVER ever should continuously run any PSU at or close to it's max output. It simply will not last or be stable. For any constant load eg miners, general rule of thumb is to load a PSU - and the line feeding it - to 80% or less of the maximum rating. So for an s5 I'd go with at least a 750w supply to keep everything happy. |
|
|
|
Got the tracking# from Bitmain for my order with Wed delivery and the last 'new in-box' s4 from Zoomhash arrives today. Of course I'll be headed out in an hour for a week long service trip  |
|
|
|
|
On the lazy miners issue, I see one of the s5's at work is in that mode right now. In the morning I'll take a few screen shots of the GUI in it's various states before I head out on a service trip to CA.
|
|
|
|
gonna have to give that a closer look. Damn good price. Thanks.
The switching outlets per-se is a non-starter of course due to the total load 8-10 miners draws but - I have a lot of contactors around with 110vac coils for brute power switching and simple relays with 110v coil that can switch the PSU control pins.
For the chronically lazy miners I'll just program it to cycle their power every 2-3 days to hopefully nip the problem in the bud. For the s5's I'll set for maybe once a week.
I only have 2 miner's on it now. Don't know what the capacity would be, but it can handle up to 15 amps which is what my electric circuit is in my home. I don't have my miner's all in one place, so I may even try to pick another one up. I just like the fact it has auto-ping and scheduling built right in. It may help users with other devices that need a good solid reboot once in a while too. I schedule my cable modem to reboot once a week, for some reason it peps right up after a reboot. One snafu I have with it is you may need to buy those 6 inch 3 prong power extensions because lots of routers and devices use power blocks that take up 2 spaces. They are very cheap on eBay. Even the 12-14 gauge ones. I'd like to find out more about the built in scripting too. I've never used it with my KNC miner but when I got my Ants, I said I may as well get more use out of it. specs say it has a 15A thermal breaker on it so that is the max total loading. The 1650w available is good for a couple s5's or 1 s4. ergo in my case the need to use it as a pilot signal for the ~14kw of server PSU's I'm feeding |
|
|
|
|
gonna have to give that a closer look. Damn good price. Thanks.
The switching outlets per-se is a non-starter of course due to the total load 8-10 miners draws but - I have a lot of contactors around with 110vac coils for brute power switching and simple relays with 110v coil that can switch the PSU control pins.
For the chronically lazy miners I'll just program it to cycle their power every 2-3 days to hopefully nip the problem in the bud. For the s5's I'll set for maybe once a week.
|
|
|
|
Yep, Gekkoscience boards are also externally triggerable for poweron/off. Enabling EON (DIP switch 3 on a full-feature board, or on by default on reduced-feature) and then give it an active-high signal (anything from 3 to 12V) on the EON header pin will kick the PSU on and off automatically. I put that feature on so they could be used easily alongside ATX supplies for GPU rails and such, and also to make remote administration easier. A single RPI's GPIO pins could easily manage a stack of PSUs from one network connection. Additionally [link=topic=902305.msg10853244#msg10853244 date=1427056080]
Just as when you change the Vcore with the advanced settings it take a hard boot - full power down - before the regulator circuits on the blades reset. Once they shut down just resetting the OS does not clear it.
Did they redesign the S5 with a software-settable voltage? Or is this a nonfunctional holdover from S4 code, which I believe does have actual chip-level regulation? As far as I know, this does not exist on the S5. Nope. s5 with its series string does not have changeable Vcore. Was refering to the s3+'s & s4, C1's there.Think it has more to do with the regulator chips themselves. Based on forensics I did for AMT with their failed A1 miners I assume the core regulators self start as soon as the rail voltage appears. From there they read their internal registers for what voltage to set. Programmable - yes. Changeable on-the-fly using SPI coms, not necessarily. Warm-restartable, nope. What's that link you gave? doesn't work. |
|
|
|
|
^^ Looking at some home automation bits to mabey trip small relays on the PSU's to cycle the rails on & off. Maybe RasPi/Beaglebone/Audrino home automation programs as well. Biggest point is low cost hardware and the ability to remote in to cycle the power as needed...
Only bad thing is that that requires poking a pinhole in Firewalls and am not to crazy about doing that. Hmm, now if M's Monitor could trip a script to a same networked power controller...
+1 to Bitmain Warrenty for clearly stating what needed to be said: Miners are exceedingly reliable but - once you start getting a lot of them odds are issues will crop up requiring the Human touch. For me it the same 5 miners that get lazy, all the others run perfectly with zero interventions, still 5 out of 37... Troubling odds.
|
|
|
|
Along the lines of when the Ants update the GUI and how...
Anyone having 1 blade of their S3+ or s5 periodically shutting down with hashrate dropping to 1/2 and yet the ASIC status show all chips are are fine? Not until a soft reboot will the status show ----- for the entire dead blade.
Do a hard boot and all is well again for several days or weeks before they get lazy again. Out of the 37 Ants in my farm 3 s3+'s and 2 s5's do this. Can't blame the PSU's because all are ran off of server PSU's feeding 2-3 miners and the other miners on the PSU are fine.
I have had it a few times on my S5. I am using a RM1000. I just have to reboot even in gui and it fixes it. If it is really happening a lot you could use M's antminer monitor and have it automated to reboot if drop in hash that far. I use M's Monitor but as I originally posted a software reboot does not restore it. Sw reboot *does* show what the issue is though, namely that one entire blade has shut down with the ----- telling that there is an issue with the Vcore. Just as when you change the Vcore with the advanced settings it take a hard boot - full power down - before the regulator circuits on the blades reset. Once they shut down just resetting the OS does not clear it. What disturbs me more is that the GUI and M's Monitor do not catch that. Both say the chips are there and working fine with the only symptom is reporting 1/2 speed. Since sw boot restarts CGminer and only then does the missing side show up, possibly presence-checking is only done from a cold boot? |
|
|
|
|
Along the lines of when the Ants update the GUI and how...
Anyone having 1 blade of their S3+ or s5 periodically shutting down with hashrate dropping to 1/2 and yet the ASIC status show all chips are are fine? Not until a soft reboot will the status show ----- for the entire dead blade.
Do a hard boot and all is well again for several days or weeks before they get lazy again. Out of the 37 Ants in my farm 3 s3+'s and 2 s5's do this. Can't blame the PSU's because all are ran off of server PSU's feeding 2-3 miners and the other miners on the PSU are fine.
|
|
|
|
S3++ would make a good case for bitmain to consider shipping just boards .... even upgrade kits for the S1 doorstops still out there!
yes, I would buy the boards to upgrade some of my S1 if they are cheap enough. Only if dirt cheap would I consider going the upgrade kit route again. Did that with 2 of my s1's > s3's and while not exactly hard to do, still a pain in the butt. My time and effort is definitely worth staying with pre-built rigs. |
|
|
|
What a load of nonsense! Absolute rubbish!
A PSU is a mining cost whichever way you look at it, I mean, they are not free. <snip>.
Yes because of long useful life a PSU can be consider as depreciating over a longer period of time but - at $20-40usd for the HP 1200w ones and only ~$35 for the IBM 2kw ones their cost is insignificant compared to miner cost. |
|
|
|
|
On that snippit Doggie spot on. I'm still using the same HP server psu's that I used to run my now retired S1's on, now they run the s3's and in 2 of them together, 3 s5's. So yes, it *is* a one time cost but they live on and on through the generations of miners.
|
|
|
|
Jeez people are buying these at close to 2 BTC, i guess people still expect bitcoin to go up in the short term ?
are these really that much better than an s4 when you factor in the fact you have to hunt for a PSU ?
Glad I ordered mine within hours, lot less than 2 BTC. Anywho, 'hunt for a psu? Should also do that to cover the possibility of the one in the s4(s) dying. So far mine are running fine but: The IBM DPS 2kw psu is perfect for an s4. My hungriest one reports in as pulling 1450-1500w from the wall... That same PSU will feed 3 stock s5's cheaply and odds are will last near forever.. Search eBay for them |
|
|
|
Also thanks everyone one for your help just glad its sorted now. (Well until i test tomorrow) I may end up buying the 2000watt ibm psu anyway, thanks for the link for the breakout boards, this was the thing i was struggling with as my soldering isn't the best Glad that went well  Ja the 2kw PSU's are great, lots of reserve margin with a S4 and enough oomph to push overclocking pretty high if desired. I'd think hitting near 3THs would not be unreasonable. and of course 2kw is perfect for 3 of the S5's @ stock speed, maybe even goosed a bit... Oh, re: soldering. When it comes to server power supplies and their associated wiring the most common problem is using a soldering iron that is too low wattage (most are 25-35w). Due to the heavy wiring and large PSU contacts being very good at sucking the heat away you need at least a 65w iron with large flat tip for fast & decent joints |
|
|
|
Hi Everyone, I think my S4 PSU died this morning, I have only had the machine 3 months or so and i think the psu is dead, could anybody help me figure out if it is and how i go about replacing the PSU I woke up this morning to find one breaker on my board had tripped, I checked the S4 and the PSU doesnt show any signs of life at all, I have checked the fuse and thats ok, but the psu does not boot at all, is there anything i can do to fault find. I know i had one of the newer PSUs that didnt have the issue the first batch did, and the S4 has been running flawlessly since i purchased it, just a bit depressing to see one of my best machines go down  Any help would be really appreciated. cheap psu can cause that, never buy a cheap psu, is the important component, you can try to return it but in the end i would suggest a corsair or seasonic 850w should be enough for the s4 850w for an s4???  Hate to tell ya this but my s4's running stock speed and Vcore draw between 1,250 and 1,375w from the wall as reported by my UPS's. My S2's report 1,039w again @ stock speeds. |
|
|
|
Hi Everyone, I think my S4 PSU died this morning, I have only had the machine 3 months or so and i think the psu is dead, could anybody help me figure out if it is and how i go about replacing the PSU I woke up this morning to find one breaker on my board had tripped, I checked the S4 and the PSU doesnt show any signs of life at all, I have checked the fuse and thats ok, but the psu does not boot at all, is there anything i can do to fault find. I know i had one of the newer PSUs that didnt have the issue the first batch did, and the S4 has been running flawlessly since i purchased it, just a bit depressing to see one of my best machines go down Any help would be really appreciated. My 2-cents is to get an IBM 2kw PSU (208-240v only) and breakout board to replace it. http://www.gekkoscience.com/products/IBM2K_board.html is a good source for the breakouts and PSU's or check eBay for the PSU's. Together they should run around $75-80 total+ shipping. I got 4 of the beasts for when my S4's croak... That's exactly what I'm doing with my 5 S4s, it's definitely the way to go. Anyone know if this would work for an S2 as well? Certainly though is hefty overkill for the S2 as they 'only' pull a hair over 1kw. If you have >208v for feed then the HP 1200w CS (Common Slot) supplies would be a better match. Just have to do some soldering with those tp attach the PCIe power leads as breakout boards are not easy to find. |
|
|
|
Hi Everyone, I think my S4 PSU died this morning, I have only had the machine 3 months or so and i think the psu is dead, could anybody help me figure out if it is and how i go about replacing the PSU I woke up this morning to find one breaker on my board had tripped, I checked the S4 and the PSU doesnt show any signs of life at all, I have checked the fuse and thats ok, but the psu does not boot at all, is there anything i can do to fault find. I know i had one of the newer PSUs that didnt have the issue the first batch did, and the S4 has been running flawlessly since i purchased it, just a bit depressing to see one of my best machines go down Any help would be really appreciated. My 2-cents is to get an IBM 2kw PSU (208-240v only) and breakout board to replace it. http://www.gekkoscience.com/products/IBM2K_board.html is a good source for the breakouts and PSU's or check eBay for the PSU's. Together they should run around $75-80 total+ shipping. I got 4 of the beasts for when my S4's croak... |
|
|
|
|
Show Posts
