GTX580 Black Screen/Reboot in Games, Windows Logo, even at Bios load

I purchased a system with dual GTX580's about a year and a half ago. I have had issues with system crashes (lockups, no blue screen, and no alt-tabbing out of them) almost from the beginning. Since I am impressed with some of the technical responses I have seen on threads in this forum, thought I'd reach out for help before I pull what hair I have left out.....

I tried literally every different bios update available, every single driver release Nvidia has released, too many clean windows installs to count, MSI Afterburner/EVGA Precision (video voltage increases, clock decreases (removing EVGA superclocking), replaced the SLI bridge, different ram, different powersupply cables, different fan speeds, and so many other things over the past 1.5 years it's rediculous.

In preparation for trashing the cards, I pulled them out and put my old cards back in. These are also EVGA brand, 8800's, also can run them in SLI with the same bridge, and not a single problem. This is while running the same driver release (clean install though) as the 580's.

So I put one 580 back in, both slots 1 & 3, and to my surprise it worked just fine in either PCI slot. Tried the other 580, also works fine when alone. Put them both in, put the bridge on, and it crashes right after Bios loads, before it even makes it to the windows logo/login screen.

Seems to have gotten worse over time (was just during intense gaming, next it's when in windows but not gaiming, sometimes at the Windows loading Logo, and sometimes even at bios load.)

Overheating? Temps are fine..even when stress testing with Kombuster/Furmark/Prime95/MEMTest/etc. temps get warm, up to 85 on CPU cores, and 75 on video card, but all tests are fine.

PSU? Bios Settings? Help?!?!?

A 1200 Watt PSU should be plenty for your setup as long as it is performing as expected. If you have had problems from the beginning with the cards, I suspect the SLI interface itself is bad on either one or both of the cards. How do they run with the bridge removed and just the cards themselves in? You mentioned running one at a time, but have you tried running with both cards in without the bridge?

Both cards run fine alone, and pass all stress tests. I put the second card back in, installed a brand new SLI bridge, and setup an MSI Afterburner profile for the fans in each card that will start with windows. I also upgraded the firmware on both SSD's, and updated the driver for my realtek ethernet and audio from the realtek site (newer than any on GigaByte's site).

So far, so good....but I have said that before.

Are you running the latest MSI Afterburner version? I know older versions had known problems with Windows. Newer versions do too, especially if they provide any overclocking via the software, but older versions cause issues regardless of how they are used.

Yes, latest and greatest.

So far, so good. Two full days without a black screen/crash. I had almost forgotten what it was like to be able to use my PC and game without crashing. Fingers crossed the this time the fix is for real!

Let us know when you are confident it is resolved. :)

I will indeed do so. Honestly, I have spent so much time on this issue, and tried so many different things, that I am seriously considering backing out my changes to validate exactly what change I might of made this past weeked that wound up fixing the issue.

Obviously the video cards are not bad, as each works fine when installed individually or together with no SLI Bridge. Likewise, since it now works the motherboard also has to be fine.

Unfortunately, the applicaiton of a new SLI bridge (the fourth different one installed) occurred at the same time I purchased a product that searches the web looking for new device drivers, and updated them, as well as me adding a custom profile and keeping Afterburner running. I should have done each alone, then I'd know for sure which one it was.

While it's possible that a new ribbon SLI Bridge worked, and the old sli ribbon bridge was bad, I would think that having tried two different three-way sli bridges (which connect both ports on top of the cards instead of one) would still be fine. Thus I don't think it's an SLI defect on the cards or a problem with the bridges, though for sure the new one I bought connects to only one 'port' on each card not two since it's a ribbon.

It is possible that it was overheating and crashing, as the cards are VERY close together and I have to think that could be an issue. Since I believe the cards are supposed to automatically change the fan speeds as they heat up, without afterburner or precision being installed, the use of afterburner at bootup with a fan profile should not be necessary. While I can clearly hear the old 8800 fans ramp up automatically as I test them, I really don't hear that without afterburner on the 580's unless afterburner is installed. Clearly one 580 goes to max gpu % on processing and the fans ramp up with afterburner, and on select tests the second card will ramp up it's own gpu % and temps will climb and it's own fan will also ramp up so I can hear it. Need to disable afterburner and stress test and see what happens. Really don't think this is the issue though, as it would sometimes crash after having been off all night and simply booting up and trying to enter windows. Certainly the cards and loads had not been on long enough to get hot.

Breaking the raid-0 on the SSD's probably had no impact, and likewise flashing the bios on the two SSD's with updates also probably had no impact. I did this for trim, which I dont' believe works when they are in raid-0 though I can't tell much difference. The Raptor's are still in raid-0.

Certainly its not a bios issue, as I am on latest version and that has not changed, nor have the settings changed (optimal settings default, then change default boot drive and enable Raid are only changes made and that pretty much stays that way at all times).

The clean install of Windows was not it, I have done that so many times I have lost count.

But what I suspect is that it was a driver issue. Since the download of the driver finder software is something new starting this past weekend, I suspect that is the fix. Previously I was just downloading all the Audio/Intel INF/Intel RapidStorage/Sata2/Sata3/Lan/USB 3/ device drivers off the vendors (GigaByte) website for the board, installing them, and then putting on the latest video driver from nVidia. None of their software, only the drivers. What I noticed was that after I let the new driver finder software find new drivers, newer than what GigaByte had available I know, installed them and rebooted, I had no more issues. Since I had already installed the Intel ones from Intel before I got the software in one of my tests, I know that was not it, but I had not done the updated USB 3 ones or the updated Realtek Lan/Sound ones.

FYI, the reason I joined SevenForums was specifically because of some of the very, very detailed diagnostic analysis and debugging suggestions I saw to other issues reported, and in particular the responses provided by yourself writhziden. Your knowledge is impressive, and your helpfulness to others noteworthy. A forum is only as strong as those who are both willing to take the time and who also have the knowledge to do so, and without doubt you are one of those rare individuals. Thank you on behalf of myself and those I see you have helped previously.

Thank you very much for the kind words. I really appreciate it.

As to why the driver update may have helped, I have a theory on that. Often, motherboard drivers for hard disk SATA/RAID controllers are too old to work with SSDs. I usually recommend getting those controller drivers directly from Intel, Marvell, or JMicron.

Its not the graphic cards overheating. While removing afterburner fan control profile, and using the built-in nVidia software's ability, the fans Do indeed speed up when the GPU temp reaches about 80 degrees, and and the cards will get up to 90 degrees, but they get no warmer than that and can run all night at that temp.

Are you still getting crashes?