Topic: The Chili – 30+GH/s BFL based Bitcoin Miner Assembly - page 16. (Read 138078 times)

Still having intermittent problems now, but I think these are "normal" ones.  Couple of questions:

- What is the recommended miner and version that seems to perform best with these Chilis (and even what flags enabled/disabled if it's known)?  I'm looking for the most stable, not necessarily the fastest; I'd rather get rid of these intermittent zombie/sick attacks.

- I'm running the 14e version with the 1.1v limitation on both chilis, but one (212 Evo cooled) still gets above 1.1v and predictably craps out sometimes.  The other is fine usually (H60 refurb water cooled).  Any ideas what I can do?  There are heatsinks on the bottom under the mosfets and above on the 4 chips above the heatsinks on the bottom.  Should I be letting some part of it keep warm somehow (cold room in winter), or is there a recommended firmware for this situation?  BTW, the Chili that is acting up has all 16 cores enabled on all 8 chips so I was really hoping the 1.1v limiation would do the trick - but alas, no.

Thanks for any help!
- Tye

I am running CGMiner 3.8.5   I also came home to one of my miners in that weird state (all lights off, zero hashing)... Rebooted the farm (with my 20 minute wait to cool down) and all was happy...

* I tested giving a miner 5 minutes of power and fans running to cool itself down enough to self test and it didn't...
* I also tested if the pi's could be turned on at the same time as the miners and would they complete their self test and start mining (even with work issued to them early) and that DID work so I can put the miners and the pi's on the same wifi controlled power switches if I want so that's good :-).


Like I mentioned before, if they are too hot to self test within 30 seconds, I have never seen them self test after that.

Thanks

What firmware on the Chilis and cgminer version?

Power cycling it will bring the draw back down to the few watt mark again, whether you wait a couple seconds or a couple hours.

On startup the board measures three sensors and waits for them to cool if needed before enabling the chips; the internal temperature of the VRM chip (30C), the temperature of the sensor near the FETs (37C) and the internal sensor of the thermal diode measurement chip (30C). Once those have dropped and settled it uses that information to calibrate the thermal diodes on the ASICs, which can vary quite a bit from chip to chip. So long as your ambient temperature isn't very close to or over 30C, it should start.

In the unlikely event that anyone else is considering trying to be creative and purchase a non-Corsair water cooler for their Chili I figured I'd relay some info about one I just picked up...

The Thermaltake Water 2.0 Pro can be made to work but it will require some field engineering! The default hardware didn't work but I was able to rig it up using everything for the 1156 mounting except the screws they provided. Right now that Chili is hashing at 33 Ghs at 64C. My other Chilis are running at 65C and 69C so am I correct in my belief that some active cooling over the mofset area will allow for more speed at this point? These things really are a tweakers delight and I'm amazed at what a difference just giving them some air makes.

Thank you for the clarification...

So any ideas why sometimes I come home with a miner with no lights on at all and CGMiner saying its connected but hashing 0?  And sometimes I'll reboot the pi and the card still wont' come back.. Power it off, wait N number of minutes (more than 5, 20 works all the time)  and it comes back to life just fine.

They shouldn't be idling at 100W after a reset, more like 10-20W the majority of which is often the fan. After a reset the board turns on the 1V power supply to the chips, but it runs at 0.85V and does not enable the clocks until the self test starts. The chips only pull a few watts at this point.

The only time I could see it idling at 100W is if it was at a high voltage after having run for awhile (like 1.15V) and for whatever reason work isn't being sent to the chips, but the clocks are still on.

well that's a minor bummer but not an end of the world..

100watts statement comes from some other comment made on the thread where idle these cards pull ~100 watts..  So if they are pulling 100 watts idle, that's extra heat that has to be removed from the card before it cools down enough to self test... (which I understand is 34c correct?)

Josh

I've found that sometimes they will get "stuck" at 36c and never quite get below that to do the self test if that 100 watts and that if I power them off, wait 20 and turn them on it works every time.

Now if no power is going to the card but the fan is running then I get MAD MAD cool quick.. (like 2 minutes).

but more to the point, is there a way from terminal/script/commandline that I can issue a "reset" or "reboot" command?

Thanks

You don't have to leave them off for them to cool down; if they're powered on while hot they'll just wait until they cool down. It goes a lot quicker if the fan is on.

Enermax Revo 1350 W has 6x 12V @ 35 A. 4 of them are connectable by PCI-E. The 5th on a different rail is from the EPS12V line so I used a "good" well enough copper diameter extension which I cutted and assembled an PCI-E as an adapter. I've never found Adapter for EPS12V -> PCI-E.
The 6th rail is on the 24pin ATX plug. I have an extension only with low copper diameter. So I will buy an good well copper one and will try assamble an PCI-E again.

Also I did this for an 1500W Enermax Platinum (4 PCI-E Rails on knc Jupiter, 5th EPS12V via adapter to one chilly. Missing one adapter for 24pin ATX)

The Enermax 1000W Platinum looks like the best for this because you don't need a special adapter. You can add 5 PCI-E cables because it has only one 12V rail for 996 W. The ones with higher rates e.g. 1500 W have 6 rails with only 35 A each.

Also I use one 550W Enermax Revolution87+ on PCI-E, one on EPS12V with 3 low performers. It could be hard on limit, but is stable working for the moment. Each single rail can support 25A, but generally not more than 45A.

Also I use one Thermaltake Smart SE 730 (3 rails 18 A on 2 PCI-E + 1EPS12V).

Edit:

In my opinion the best is to use an mono rail power supply because you lost to much A on different rails or the rails did not have enough A to support 2 Modules on one rail :-(

I have an very old Enermax Noisetaker for 600W. It supports 2x18 A, one on PCI-E and one on EPS12V.
I use this for tests of my boards.

I guess your PSUs have no output electronic fuse which emergency shut down the psu. I killed also one old style psu :-/ which was allready in my electronic trash bin. Now it's back in the bin but without the internal cooler which is still usuable. I recommend to us an psu which has no problems with overloading an save themselve. I've allready two defects on enermax psus but never there destroyed my connected hardware.. So I feel to trust them. But I think also there are better ones on the market ;-)

So I have noticed,   when my cards go "south" I have no way to remotely reset them... I have to turn them off, wait for the chips to cool enough to self test ( usually 20 minutes) then turn them back on.  And often times, I have to do this with my Pi's off, so that they don't issue work to the "off" miner while its self testing.

Has anyone come up with a way to do this remotely?   I have the opportunity to move my mining to my work, but before I do that, I need to understand if there are any terminal or other ways to force the miners to cool down, and truly "reboot" themselves remotely.

Thanks all!

Does anyone know why my Chili would be running at half speed? I can only get 16GH/s out of it: https://i.imgur.com/olezO7v.jpg

Thanks

Does anyone know why my Chili would be running at half speed? I can only get 16GH/s out of it: https://i.imgur.com/olezO7v.jpg

Thanks