I've been dealing with this issue for a couple of weeks now.. and I'm hoping you guys can help. I've been pouring over these forums, and may have it narrowed down.. but still not sure.
The problem. Random shutdowns. Sometimes during gaming, sometimes during idle (over night). I'll wake up and everything is off except for a blue light on the mobo. Computer wont start back up.. it'll power for a moment, but will not post. Have to disconnect the 24pin power, or leave off and unpowered for a bit to get it to power.
Sometimes it will boot fine from here, other times, it will say cmos checksum error, and require defaulting the bios.
mobo: EVGA X58 sli LE (POSSIBLE FLAG? <== I was told the board was powered on with a usb connection in the 1394(firewire) slot, which cased the 1394 chip to start smoking.. there's a little scorch mark on it. Though, I never had problems until almost a year later..but red flag all the same==> )
cpu: D0 i7-920
gpu: EVGA gtx580 (x2 sli)
ram: corsair xms3 1600 12gb (2gb x 6)
cooling: corsair h100
psu: corsair hx1050
win 7 64bit pro
wd velociraptor (boot drive), wd cav black (storage) ocz ssd (fraps vid cap drive)
ok.. so I built this system about a year and a half ago. All parts were new, with the exception of the motherboard and cpu, which I received in a trade. It's my first actual build, so yeah,. guess that doesn't matter.. but I'm proud all the same.
The first thing I did was experiment with GPU overclocking. After some trial and error, I settled on 877 clock for my daily driver, and 907 as my God-I-need-my-fps-at-120-to-show-off profile. The cards run at 1.088v and when gaming or benching.. stick around 65-70 steady. I clean them regularly.
After some months, I decided to give a try at cpu overclocking, something I had meant to do for awhile. Lots of research, trial and error, and some stability testing, I hit 4.2ghz, and was stable under prime95 for about 24hours. Yay me. This is about the time the problems started...
the start of it
Gaming away in some Aion, enjoying my 120fps in town for once, I get the first shutdown. Temps seemed to have been fine while gaming, as I keep monitoring software running on another screen. No blue screen, no slowdowns, lag or artifacts.. just "click" and the pc was off. Pressing the power button did nothing. Cut the psu power off and waited for it to discharge, and I was stuck in a boot loop. It would power on for a second, and you could hear the fans start to spin up.. it would do this like 3 times quickly, vrr, vrrr, vrrrr, pewww. Ok.. so I can't really type the sound out, but after trying to boot, it would just shut off. No beep, no error code on the motherboards readout.
I pulled everything off the board, reassembled and it came up. ok.. little worried. A Few days later.. I come home from work and it's off, except for the blue status light next to the 24pin power connector.
This has happened several times now.
what I've tried
So, I've been pouring over forums, websites and manuals. The first thing I did was check the event viewer. System was shutdown unexpectedly. >_< really? Thanks. But I guess it shows that it must've been a hardware issue, as windows didn't detect something and shutdown because of it (maybe I'm wrong here).
Memory. maybe? I test each stick individually, with memtest and each passes. I've tried running the computer with only one gpu (swapped out both) and neither one made it a full 24hours before there was a shutdown. I've cleaned every connection I could, have defaulted the bios. Tried stock memory voltage and timings, tried xms profile 1, tried setting the timings and voltage manually. Still get the issue.
PSU. I didn't want to believe it could be the cause, it's a good brand (so I'm told) it's higher end and isn't that old (imo), but.. maybe.
I come across the program AIDA64, and give that a check. +12v shows at between 11.7 and 12.03 at idle, I check with evga's eleet. Just about the same, except for eleet seems to update faster. Then I check on stress with AIDA64. the +12v drops to about 11.6 at the lowest.. ok.. not good I think. Then I try running evga precision's burn in gpu oc test. Running stock on the card, the voltage showed as low as 11.15 on both AIDA and eleet. Eleet would show it first, then AIDA64 would match. Both prog's show exactly the same.
Now, I hear (read?) that software voltage can show off.. so I go dig in the garage and find a DMM. Not really sure on how to use it properly, so I google. One site said to use a free molex connecter. I shutdown, plug an extra molex into the psu,.. plug in to the ground and +12v and on powerup I get 12.06 steady. I start running the cards, and nothing changes on the DMM, but both software again show a drop to 11.15 volts on the +12 which stressing a single card.
I have a suspicion that running 2 gpu's OC'd with and OC'd cpu, black lights (yeah.. I have black lights in the case)
and the h100 water cooler, may have pushed a slightly defective psu too far, and now I'm getting shutdowns related to the +12v dropping to 11.15v or lower.
I'm not sure what else to test, but I'd like to verify with my DMM, but testing on a free molex isn't showing any fluctuation in voltage, while two separate monitoring programs show the exact same drop, that is easily reproducible.
I have the receipt, boxes and whatnot, and I hear Corsair has great service along with a 7 year warranty, but I would like to confirm before contacting them.
Is there a better way to use my DMM to check the voltage under load? Where do I check? On the 24pin? the 8pin for the board?
That does sound like the PSU. Though Corsair are usually pretty good, but strange. Doubt it's anything to do with the CPU OCing.
Given that the DMM says voltage is fine but the board disagrees, check the 24pin is plugged in properly. Sounds like a bad connection.
I've unplugged and re plugged the 24pin a few times now. I think I've got it on there pretty well, though tomorrow I'll examine it a bit more thoroughly.
Would checking voltage with a DMM on an empty molex connecter be the same as checking say.. the 8pin plug for the mobo, or the pcie one for the gpu? It seems odd to me, that the software shows a drop in voltage directly related to stressing the cpu and gpu (larger drop with gpu) and yet no change on the DMM. Should I get a reading from somewhere else using my DMM?