OK Everything's running fine. I've been using Rivatuner to OC my GTXs and GPU temps coming out of 3DMark06 back to desktop are high 60s low 70s which should be fine. Well I unlocked the shader OC option in Rivatuner 2.05. I bumped it up 200. After reading on the SLI forum where there are other users who have OC'd the GTXs more than me and are running stable ... so they say.

Well I run 3DMARK06 again. At the test with the blimp and water dragon, it starts to run choppy and I get some black blotches on the screen. I figure the OC is too much and reset the PC with the intent to back down to where I had it before.

At boot, the PC starts to load then locks up at the Windows boot screen. I reset, and read the 2-digit LED readout. This time, it stops at post code 75. Not having my manual handy, I figure I fried my GTXs? Pissed, I open the case and look at everything. The PC then continues after a long pause and the NVIDIA boot agent appears. Says cable may be disconnected or something or other. I never had this problem before. I took out one GTX so there's no more SLI. I just left it and went out to dinner.

About an hour and a half later I come home and boot again. This time it boots into Windows fine. Wierd. Anyway I run 3DMARK06 at the stock freequencies, disabling the OC completely on the GTXs. It ran fine, and I scored what I did some time ago before the OC .. 11838. So I left it on, at the desktop to go do something out of the room. I come back, and it locks up on my 3D screensaver. The screensaver sound is looping.

I reboot, and get post code 75 again. I swapped GTXs and left it along for a few hours. I come back, and it boots into Windows fine again. I play the Crysis demo to see if any artifacts are present. Again, none of the GPUs are OC'd. The demo plays fine, but then I get these really long pauses randomly. It'll play fine for a few mins, then I get a pause for one min, then plays fine for 10 mins, then a long pause, which this time seems like a lockup.

I reboot, and post code 75 comes back. I left the PC alone for 20 mins, and start it up. Code 75 still shows up.

This morning, I fired up the PC after being off all night. It went right into Windows. I had to rush, so I just shut it down again. It seems this problem occurs after the system has warmed up.

Now today at work after doing a search, it seems this post code has to do with detecting the HDDs or devices. Not video related? I also read that disabling onboard LAN or wake-on-LAN will fix this. Strange, I thought I had wake-on-LAN disabled. I'll try these when I get home. I also have an older GeForce4 card taken out of my bro's computer. When this happens again tonight I'll swap the cards and see if I get the same results. If so (which could be good) it'll rule out the fried GTXs, which was rather pricey.

I guess it would lead to the mobo or RAM? I'll also run memtest. I'll undo all OCs in BIOS as well after trying everything else. Maybe that'll help.

Maybe my story of last night will shed some light:

I tried to mess with some Power User settings in RT to get it to OC my 169.04 driver, and I must have messed something up because I started to get REALLY bad frames in 3dm. So, I uninstalled RT. Still bad frames. So, I installed 163.69 again. Still bad frames. Finally, I decided to uninstall/delete the driver from Device Manager and check - from safe mode - that all the NV display driver files were, in fact, eradicated. After reinstalling 169.04, I was back to normal.

My point: did you try uninstalling RivaTuner (and deleting any residual files)? AND, did you uninstall the driver completely (manually deleting residual files)?

No I didn't uninstall anything. But those are at the driver level, right? Doesn't that mean they come into effect after Windows boots? My guess is I have a hardware problem somewhere as it is initiated during post. When I was able to go back into Windows and ran 3DM, I did not see any performance loss from my scores.

Upon further research I've read the MOSFET may be problematic now on the mobo too. So many directions to take, but I'll try and diagnose one thing at a time.

Which mosfet? There are many.

Driver-level changes sometimes can affect hardware... I'm thinking that because you did something in software to cause a hardware error, maybe you can undo it at the driver-level. It sounds stupid, but I'd give it a try.

Another thing: try resetting CMOS and re-flashing BIOS.

I think there's an updated BIOS P31 out right now, I have P30. I'll update the BIOS and see what happens. I think my current attack plan is the following:

1) Boot into Windows,
2) Download GeForce4 drivers for XP**
3) Uninstall RivaTuner (only used to set fan speed at 100% currently)
4) Run memtest86.
5) Run 3DMARK06
6) Play Crysis until crashes, or for an hour (assuming I'll get a crash)
7) Shutdown PC, replace GTX with Geforce4 card.
8) See if I still get 75 post.

** If I get the 75 post at all, I'll just go to steps 7 and 8 right away. If I still get this error, I'll know the GTXs have nothing to do with the problem.

I will uninstall the drivers and install the most current, see if that helps. If not, I'll reset the CMOS.

Anything I'm missing?

OK I think I found the problem ... but the result is wierd.

I opened the case again and checked everthing. Powered up, booted to Windows. Then programs started to stop responding. I rebooted and it went back into Windows. I ran memtest windows version and the system locked up. I figured it's the memory. Rebooted, gives me the 75 code.

BUT, just to be sure, I uninstalled the GTX and put the Geforce 7 in (not GF4). It requires no molex adapters, just the PCI-16 slot. I also checked all the connections to the HDDs and DVD drive, both SATA and power. They seemed secure, one on the mobo may have been a little loose. Anyway I boot up again and it doesn't stop at code 75. Nice.

I start up in safe mode and run memtest. All four memtest windows (allocated 500 each, except for the last one, which is for remaining unused memory) ran to 100% without a lockup or problem. Thinking that one of the cables was just loose, I took out the GF7 and put the GTX back in and went into safe mode. Did same memtest procedure, but it locked up within minutes!!

Now I was REAL confused, as it had to be either the card or the PSU. Trying to think of something else to test, I connected the other two PCI 6-pin molexes that would have gone to the secondary GTX card for SLI.

Rebooted into safe mode and did same memtest procedure. Now ... it ran to 100% and beyond. I swapped out GTXs and did the same thing. First two 6-pin molexes locked it up, third and fourth did not lock it up at all. But the kicker is that all of the molexes were connected to the PSU fine.

I ran the Crysis demo for almost an hour, not lockups, artifacts, or long pauses. Then ran 3DMARK and got a score of 11953. No problems.

Here's the kicker ... my PSU is only one 12volt rail. So how could this have happened???!!!

I'm going to try a few things tomorrow and see if I can remedy this problem.

Back again after 6 months. I got frustrated and only used my gaming PC for email (before it locks up & resets) and my old Dell laptop to surf the net and all other computer related stuff (bought a house too, took up lots of time and $$)

Anyway, I now have time again and the money to potentially upgrade. But I'd like to know more about what the problem is here, first. Maybe I can get some experts to chime in:

1) Using my old GeForce4 card (no molex adapters), the system still locks up after use, so it's not the GTXs that are causing it. It takes slightly longer to lock up maybe with the GF4?
2) I noticed that if I take the side off the case, the PC stays Ok longer (5 - 30 mins average). With the case on, it posts the 75 code more frequently
3) On humid days, taking the side off the PC doesn't really help much. After running the central air in my house to get rid of the humidity, the PC posts OK a lot more frequently.
4) There were some days where I could play Oblivion and the PC wouldn't give my ANY trouble, then it's back to square one again with the 75 code.

My guess is something in the PC is now very sensitive to heat.

Is there some how I could just test the PSU, or mobo, or RAM, or CPU independantly without putting each in another system? I could just plan on building another PC and replace one component at a time to see what happens, but I want to wait a bit more before doing that if possible.