Bitcoin Forum

I've put together a rig consisting of 4x 7950s in a milkcrate with powered risers running on BAMT r9 1.3 from USB drive. 1000w EVGA PS pulling 1100 at the wall.  This thing is currently keeping my garage above freezing but it is not as stable as i'd like. 

it seems to work great most of the time but occassionally i'll notice a card shuts down for no apparent reason, or my CPU usage pegs and the system is unresponsive. Last night I had 1 card that kept throttling back and reving its fan up as it was banging around 70c when the other cards were mid to low 60s. I can usually issue a coldreboot and it'll come back up and work fine for a while again. Sometimes after coldreboot it comes back up, i can ping it but no Remote Desktop availability. I have to Putty in and do another coldreboot or go physically reset the machine.

nothing is overclocked and BAMT seems to be adjusting voltage on the cards while running 2.2-2.3 Mh using the default config.  I'd like to get this thing stable then start tweaking the voltages down adjusting hashes/card... Is PS likely the problem? It seems to do alright so long as i just let it go but if i start trying to switch pools or am constantly monitoring the hashrates is when i notice issues.  BAMT r9 1.3

could a flakey SD drive be the issue?

Are you using a USB or a SD drive?

I know, I was wondering why flakey was saying if a SD drive would be the issue if flakey was using a USB drive.

I'm thinking you are asking too much power to the psu it becomes unstable at some point. I'm not really sure a 1000w psu can handle 4 gpus and the rest of the rig, but if it can it would be with some undervolting.

Psu's are not meant to be used at full capacity 24/7, you should either leave some room or modify its cooling. I would bet that your psu is running hot.

Are you using Windows? If so, install cgwatcher

7950's are 200W cards, and 1.1kW from the wall = 880W from a 80+ power supply, like that evga. Seems about perfect to me, even within 80% max load like I'd recommend. What is it you were saying again?

Fans revving and randomly going dead while mining sounds like a riser issue to me. When the cards die, is it usually the same card in cgminer? You've shuffled them (the cards) around a bit already, right?

Serious?

If you people can't read, don't freaking post. You aren't doing anyone any good.

You are assuming the OP uses 120v and not 220v. If he is using 220 he is using almost 1kW from the psu. Also I've read each 7950 draws 225w on stock volts, x4=900w. If he has a cheap 60w cpu and only 2 slots of ram filled he is at 990w. If he is using some good cpu then he is already over the psu maximum capacity. A good psu will shut off if you demand more than it can handle.

Hi all, I have a digitalOCean droplet running. does any one know how many workers to put on it for the optimal performance.

Thanks

You should create another thread for this...

Just use one worker. Having multiple workers mining on the same machine doesn't make much sense.

1100W from the wall is 1100W, period. All else is irrelevant, no matter what you've "read." Do you even know what we are discussing here?

If he is using 240V, and that's a long shot, the PSU should be more like 85-90% efficient. That would still be 990W at the very most, which I know that power supply is more than capable of doing. If you are so sure he is overloading the power supply, he would see a hell of a lot more than 1100W at the wall. Of course you know power supply efficiency drops a lot when very near the rated capacity, right  ::)

I think we are discussing possible reasons of an unstable system. You might be right on the wattage of the cards, but I could still be right about the psu being pushed to its limits.

OP has not stated how much ram and what cpu it has. Their wattage could vary from 75w to 200w easily.

He hasn't as well told us how he measured the the current at the wall. Clamp meters can be innacurate if used incorrectly. He might have used a kill-a-watt though.

220v is not that far-fetched, most of the world has 220v.

And if he managed to position his psu in a way it absorbs some heat from the gpus, the psu might overheat and produce some voltage drops. Since psu its working in between its limits the protection doesn't trigger and the rig remains on.

Other than that the only things that is coming to mind for the rig to become unresponsive would be the ram. A bad ram or insufficient ram making some of the cards produce hardware errors that could make it unresponsive or make cgminer call a gpu sick.

Don't get me wrong, I'm not here to argue with you, but in my experience, when troubleshooting some issue, it is always better to keep an open mind about it  ;)

We still need more information to be certain on what the problem is. But I don't think my answer was that crazy though...

sorry i haven't come back to this for a while... Power was measured at the wall with a kill-a-watt, 120v. System is running rock solid since moving to Litecoin-BAMT 1.4. I think there must have been some driver issues causing the instabilities on the original BAMT install