Topic: Top-Tier Rig Random Shutdowns - page 2. (Read 5018 times)

Ah shoot, it overheated and shutdown after 45 minutes of mining with the small fan to help move air along, as opposed to 5 minutes without the small fan. Well this definitely suggests to me that this is a temperature issue. :/

I've been looking through Newegg, and this case: http://www.newegg.com/Product/Product.aspx?Item=N82E16811112159 looks like the case I'm going to order. Plenty of fan slots, top-mounted PSU, 10 slots (so I can use a couple risers to give a 2 whole slots of space inbetween 3 cards..!) And I'm planning to replace all 4 of the 140mm fans with these: http://www.newegg.com/Product/Product.aspx?Item=N82E16835553003 .

I'm also going to warranty-replace the PSU while I'm in the mood for ordering new parts, for good measure.

Further ideas/suggestions/comments welcome.

lol, yeah I've kinda suspected for a couple days now that the real issue might be that the case is just too small and the PSU might actually be overheating, yeah. And that caution sticker is the warranty-destroying sticker over the GPU BIOS switch. I'm leaving it on so that I can keep the warranty in-tact, but I'm using a small (small!) screwdriver to adjust the BIOS switch as I see fit.

I think I can still get away with this setup though. Assuming that power isn't an issue (which I'll figure out Tuesday when the Kill-A-Watt arrives from Amazon), and the PSU really is just overheating, then my "fix" might actually work here. If the PSU is overheating because its HUGE fan isn't moving enough air through the PSU, why not just increase airflow through the PSU with an external fan? So right now I have a small fan pressed against the power supply grill on the outside, and it is moving a LOT of air through the PSU. I can feel it sucking a lot of warm-hot hair out of the PSU.

So interestingly enough, before whenever I would try to have all three cards running simultaneously mining, and the cards were stable temperature and clock wise and they didn't crash (which would reduce power usage since the GPU would no longer be mining) then the whole rig would shutdown after ~4-5 minutes. With this little fan on the external side of the PSU pulling more air out, I've been running the whole rig mining with all cards for at least 20 minutes now, no issues. Here's what I'm currently looking at right now over VNC:



And I think the clocks are @ 935, 970, 925, 830, 830 right now. Adapter 1 is a bit warm, so I could probably reduce it by 10MHz or so, and the bottom 6990 (Adapter 3, 4) is hanging around mid-80's, so I could probably increase it to 850.

Assuming the PSU doesn't overheat, I think I could actually get away with this.

Hooh my GOD... dude, that case is WHAY TOO FUCKING SMALL for those cards...

And yeah! Your second PSU should be powering 2 of the cards and a hard drive (to provide a 3.3v/5v load), not the mobo or anything else! It MUST start up at the same time (or shortly after) your M/B so the cards don't get brainfucked by the lack of a M/B power load or bus clocks (and so they don't back-power the board itself).

1: are you fucking serious, take that bottom card out. NOW. That's why your power supply is shutting down, it's NOT GETTING ANY AIR around that video card that's directly blocking the PSU fan...
2: "(!) Read the..." (and likely "remove this" somewhere on there too) - ummm.... did you read that tag? o_O
3: please ship me your rig so I can do it right >.<

ACtually looking at your pictures again, doh, yeah your case is too small:



clearance between the bottom 6990 and the PSU is really not enough since the fans are fighting each other, and theys suck in each others heat. Particularly with the side panel on.

Im assuming you wont run this noisy space heater in your study, but somehwere in the basement or wherever, so Id just leave the side panel off and pull the PSU out and put it next to the case. Problem solved most likely. If you must fit everything in a closed case the above advise of turning the PSU upside down and cutting a hole in the bottom is a good one (if you give the case some bottom clearance), but honestly you want a bigger case, or none at all

NLA,

Man, this has been an epic saga getting your rig to work! Hang tough, man. Just wanted to say that before you buy a new case for top mounting the PSU, you might try just turning it upside down so it is pulling in cool air from underneath. Since you have an oversize PSU, you might have to cut a hole for the intake, but you might have to do this for a top mount case too (I'm really not sure to what extent cases are designed for oversize 1500 watt PSUs). Sorry if someone already suggested this - haven't had time to read through every post. After my experience with mining in a case, I would just forget about the case and go with an open air  rig. I realize this is probably really unappealing or simply not an option for you, but cases are such a hassle when you are running 3 or more GPUs at full load. I started with two 5850s in a case with great airflow, and even then cooling was an issue. So when I expanded to four, I lost the case and haven't looked back. And I've grown fond of this funky industrial tech art object sitting in my house, lol.

just a thought...
try hooking up the 550PS to your motherboard...
that will then power the board and save on using power from the 1500W PS
leave the 1500PS plugged into the harddrive...

and plug in one of the 6970 GPUs to the 550W PS ,
or perhaps split the load between the 2 PS where only 1 PCIE power connector is in the 550W PS and the other comes from the 1500W

myself i'm running out of ideas...

I have noticed on occasion my 1500W PS on shut down refuse to start back up... this was on shut down for maintanace so the power down was for a short time...
then afterwards it would power back up just fine... (heat of the summer and the room was pushing 37-39 degrees...

Nope, I only took the side panel off to take the pictures. Also, leaving the side panel on provides a lot of extra cooling to the top 3 card slots, since there's an 80mm fan mounted in the side panel that moves a lot of air onto the cards.

And yeah, the extra molex is attached.


Like I said, the PSU would just shutdown randomly unless I fiddled with the paperclip again. Something wasn't working right, which is why vapourminer was suggesting to plug in a few hard drives. Which I'll probably do later tonight after I've left the rig mining with 3 cards a for a good long while, monitor the temps, etc.

Still open to suggestions/observations!

EDIT: It's been running for 1.5 hours at full load. Temps for the 3 GPU's running (all 6970's) are <= 80C. Good stuff. I ordered a Kill-A-Watt from Amazon, should arrive Tuesday, will give me a good idea of how much electricity is actually being used. If the system is crashing @ <1400-1500W of usage, either the PSU is defective (and needs to be replaced under the 3-year warranty) or the 6990 @ 100% fan speed is stealing too much airflow from the PSU and causing it to overheat.

question...
are you leaving your side pannel off?
reason i'm asking is the 6970's will dump all its heat out the back...
but the 6990 will dump its heat out the back and the front.

but i don't think your powersupply is over heating...

also does your mother board have an extra molex connection for power?
the PCIE bus will supply 75W of power to each video card...
thats 300W alone that the motherboard will supply....
mind you i don't know if the video cards pull that much power from the PCIE when they have extra power supplied via cables.
just a thought...

did u get a chance to hook up one 6970 video card to your 550W psu?
if so how did that go?

Sure, will do. Gimme a few minutes.



EDIT: Alright, I'm running the top 3 cards right now at full load, with the bottom 6990 sitting idle and with its fan set to a lower speed. Current clocks: 900/180, 950/180, 880/180, yielding a lowly 1220MH/s. :`(
Typically when the whole thing crashes, I can feel the metal plate on the back of the PSU feel REALLY hot. Right now it only feels warm. We'll see! If it turns out that this is a general problem for bottom-mounted PSUs.. well, I'll just kick myself and order a new case with a top-mounted PSU I suppose.

Post a picture of your setup...
this will give us a better idea of how your cards affect everything else

ahh. my apologies to you then.

however it was right after my post and you hadnt mentioned anything about it, I figured you didnt believe the post.

Can you first let us know if the rig is stable with 3 cards?

As for the PSU overheating, I kinda doubt it, but you could just pull it out and find out.

No, read my question. I'm talking about something completely different.

I understand the likelihood that I need to plug in something in order for the power supply to function properly, powering a 6990 or 6970 by itself. But those pictures don't have my 520W in it. My question goes back to the 1500W PSU in the case, which is shown in the pictures. The 1500W has shutdown randomly for full-load overclocked and stock speed GPU configuation. My question has nothing to do with the 520W supply right now.

I'm wondering if the 6990 is robbing enough airflow from the bottom-mounted 1500W supply to cause it to overheat. 

read this again.

YOU NEED A MINIMUM DRAW on 5v and 3.3v or it WILL NOT operate correctly.

sorry for the bolding but that is your answer.

http://imgur.com/a/hBhuI#7oxAx

Any chance the low clearance between 6990 and 1500W PSU and the opposing airflows might have something to do with random shutdowns?

that "min" column... that is minimum draw needed for it to operate correctly. thats what Im talking about. you need 1 amp on 5 v and 0.3 on 3.3 volt.

without that it will not regulate properly, and when you turn on the other psu something goes over/under minimum and its protection kicks in.

try plugging a couple hds in it and try again.

So, where does one learn apparently elementary things like "ground it with a paperclip to start the PSU?"

I was thinking about building a rig myself, but decided not to because I am concerned about the viability of the bitcoin market.  This thread is making me glad I decided not to simply because my knowledge of electronics is so rudimentary.

I see information in the manual about minimum power supplied, but nothing about minimum power required. And its not too old either, I only bought it in '07.
http://www.newegg.com/Product/Product.aspx?Item=N82E16817121013

Edit: I'm wondering if maybe the way the airflow of the rig is set up is leading to a robbing of airflow through the 1500W PSU. The rig has 4 video cards in it: 3x 6970's on top, 1x 6990 on the bottom. The 6990 intake fan sucks up air in order to cool it off. I have all the fans of the video cards running @ 100%. The 6990 intake fan is directly above the PSU intake fan. I wonder if this is messing with the flow of cool air through the PSU, and leading to overheating and random shutdowns? Will post pics in a little while.

that kingwin psu. does it have minimum power requirements on each rail? like you must have 1 amp on 5v? many older psus will not run stably unless you meet minimum draws too. check the label on the psu.

to get around that some folks plug old hds or opticals in.

You can plug 6 pins PCIE connector in to a 8 pins "socket". A 6 pin is rated for 150W. x2 =300W. Add to that 75W from the motherboard, and it should be plenty for a 6970 at stock speeds.

You havent mentioned though, if you plug out at least one card, does the system work fine under load?