What Broke?

[d0gr0ck]

Hey guys, I'm hoping someone out there can toss some ideas at me as I troubleshoot what happened to my PC.

The PC specs:
Q9550 (was OC @ 3.62GHz, stock for troubleshooting) | Asus P5E3 Deluxe Wifi-AP@n | 2 x 1GB & 2 x 2GB DDR3 1333 (currently at 1066 per Mobo defaults) | 2 x 4870
There are some other bits, but I do not think they are relevant to this case.

The Story:
While playing Mass Effect 2 the other day I suddenly had lockups out of what was thought to be a stable system. First time I though was a fluke, caused by the game. After a few times I dig into the error reports and find that the ATI graphics driver is at fault; logical step: reinstall the drivers (Catalyst 10.8). I start playing games again the next day and everything goes blank but the screen still has a video signal. Hard reset and nothing but fan noise.

My PC is no longer even getting to BIOS. After removing the coin cell battery, clearing CMOS and only leaving one stick of RAM in I get it to boot. However, the system hangs during POST randomly. The screen cuts out but there is still video signal; in fact, this is the same symptom that first showed up. This also occurred as Windows was booting.

After some patience and a BIOS flash I can log-on, but the blank screen can happen at any moment. Sometimes after a forced restart I have to do a full power cycle to get the system to POST.

Troubleshooting so far:
Dumping any and all overclocked settings: no change
Full graphics driver sweep and re-install: no change
Removing Crossfire bridge and using primary GPU: no change
Swapping around the RAM: in progress, no changes so far
Test PSU voltage: in spec
CHKDSK to make sure Windows is good: no change

I'm slowly moving to wit's end here and am seriously dreading that the mobo has up and died on me. I've got a DX48BT2 lying around to transplant everything to, but with the odd graphics error reports I don't know if that will solve anything.

It's 3am now so I'll be back on this in the afternoon and trying any suggestions after some sleep

 

[Motopsychojdn]

I'd suspect your psu as a first offender,
do you have another you can swap out to check?
Moto

 

[4745454b]

You said you used primary GPU with no changes, have you tried using the other GPU?

By the way, I had the DDR2 version of that board, your not supposed to use CF on it. 16x / 4x CF on PCIe 1.0 is a no no. You get the performance of 4x/4x, which is usually slower then a single 16x.

 

[d0gr0ck]

>Moto
I'd really hope not. The PSU is a TX 750 that was already replaced over the summer because it started to fail. There's a guy down the hall from me waiting for his replacement HX 750 (capacitor blew) who might be able to lend it to me for a bit when it arrives. Until then all I know is that the +5V and +12V lines show nothing out of the ordinary on a multimeter.

>4745
Just attempted to start the system on the Second GPU, no video signal. Odd since merely removing the crossfire bridge and changing the monitor cable before boot has allowed me to use it as primary before. Are you sure you're thinking of the correct P5E3? Mine is quite clearly an X38 chipset with 32 PCIe 2.0 lanes split evenly over two full length slots.

Progress is slow owing to the fact that there's a 50/50 chance that POST won't occur. Swapping GPUs between slots is a challenge due to them being tightly bound in my LCS. It should not matter if you use a GPU from PCIex16 slot one or two and hasn't in the past.

I've currently unplugged the power from GPU 2 but left it in it's slot (have to drain LCS to remove it). After anther full power cycle I'm in Windows and waiting to see what will happen.

Edit: Sonofva...! As I made this post the screen went blank again but still had video signal. Back to the drawing board...

 

[d0gr0ck]

Quick update: I managed a few safe mode boots and swept out any and all display drivers using Driver Sweeper. I'm now booting into Windows with good reliability, but not always posting. Once in Windows everything is stable; going to reinstall display drivers later. After that I'll try and setup crossfire again (my second vid card is unplugged at the moment). For now I'm gonna see if I can pull anything useful from the minidumps folder.

 

[4745454b]

Obviously I am thinking of a different motherboard. I had a P5E-deluxe with wifi and knew there was a DDR3 model of that board. Same board but with a DDR3 controller instead. Mine was a P35 board, as was the DDR3 model. (NM, just dug the board out, I'm thinking of the P5K/P5K3)

Seeing as you've lost one PSU already and things seem to be better with a lower load/one GPU I'm not willing to rule that out yet. Sounds like your living a dorm/apt building so the power delivered might not be that conditioned/good. I would try borrowing his when it arrives and see if its better. If it is, I'd get a UPS so this doesn't happen again.

 

[d0gr0ck]

So yeah... I was about to say how it seems like the whole problem was with my second 4870. I popped it out and everything was fine and dandy for a few hours. Just as I was about to have a quick game of L4D2 the whole thing goes tits up.

Automatic restart and new BIOS beep code: No VGA detected. My remaining 4870 now has a little red light on it, nothing good comes from little red lights. Cycling the power everything boots successfully. The minidump indicates atimdag.sys is at fault (divide by zero error to be precise). I'm going to flush out and re-install Catalyst again.

PSU does seam more likely as the odds of two GPUs plus drivers going at once is just kind of out there. I live in a decommissioned frat house (boarding house now really) from the 50's so power conditioning is crap. Though the last PSU failed over the summer while living at home where the power is clean. I've only got 2 20amp breakers for the entire room and despite best efforts there is a mini-fridge on the same breaker as my PC. I can't seem to get the combination of outlets to have fridge, PC and sound system all separate.

I've been invited to a have a beer and watch a movie now, and I think I could really use it. See you tomorrow!

 

[4745454b]

Good luck. If the power is bad I would get a UPS. You might need the new PSU as well, I'm not sure and only testing will tell.

 

[d0gr0ck]

Picked up a new PSU today cause I got bored of waiting. Who woulda thunk it; my local Best Buy had a single Corsair TX 850 sitting on the shelf a couple dollars cheaper than Newegg.

Yes the old PSU is definitely on the fritz. Gave the system a quick linpack run and christened it with some NFS Hot Pursuit. Hardly conclusive, but the ability to run a game is already a major step.

The jury is still out on the second 4870, I hooked it up again and the screen went all pixel junk colors the moment a load was applied. Probably going to RMA the old PSU and sell the one I get back. I'll also need to see what VisionTek will do about honoring their lifetime warranty on an older 4870. I'll also be making a trip to Home Depot to to find some surge protection so I can avoid this scenario happening again in a few months.

Thanks for all your help guys!

Edit: Wow, it's a nice little mess around my PC. Perhaps someone will be amused by the odd state of things:

 

[Motopsychojdn]

/passes out at sight of cables on show
Moto