I offer the gift of love for whomever can help me solve this problem.

Tellox

So, long story short - I saved up for a long while, and could finally build what I dubbed my dream PC. The setup was;

Asus p9x79 Pro motherboard
i7 3930K cpu
32GB DDR3 (Ripjawz 1600mhz, 8x 4GB)
H80 Cooling
2x GTX590 SLI
Corsair 240GB 6gbit/s SSD
2TB WD
Corsair AX1200 PSU
Thermaltake Level10 casing

Now, on the initial build, I left out 4 sticks of RAM - as when I mounted the H80, I realised I'd have to take it back out to put the RAM into the slots behind it.. and well, I was getting pretty tired.

I connected up everything else, and away we went. First problem I noticed was the mounted fan I put over the RAM (A Geil Cyclone) was only running at around 2000rpm, wheras it's meant to run closer to 4000. This isn't a fan that kicks up when it's under load, it's just a constant speed - Nevertheless, the temp was reading at about 18' C, so I wasn't too bothered about it.

Everything else went fine - after a bit of wiggling around, we passed POST and we were raring to go. Unfortunately, the problems started shortly after. Once both Windows and Drivers were installed, I of course, began overclocking. I used the Asus AI Suite to do this for me, and got up to 4.7ghz out of the cpu. I probably could have gotton it higher if I did it manually, but at the same time this was more than suitable for me - and I was looking to start the benchmarking process, of course.

The system started locking up around this time - sometimes during a benchmark, sometimes when the PC was on idle. Also some odd glitches such as pink screens, graphical issues, etc etc. First thoughts being graphics card, I removed one of the GTX590's, and things started going smoothly. Rang dabs and they were happy to refund it.

Now, from there, the problems continued in gaming, or when the machine was under load - often during a game of starcraft2 (or anything that was pushing the card), the display would stop receiving a signal, and the audio would start skipping, until a hard reboot was done. Event log was showing Kernel-Power as the issue. One, the PC managed to recover - and displayed a warning stating that the Nvidia kernel had stopped responding, and was recovered. So, I assumed, another ****ed graphics card. With this, whenever the system crashed, it would show up with "Overclocking failed!" during boot, forcing me to re-do the UEFI Process. After a few of these crashes, I set the CPU back to it's stock speed, incase that was the issue.

I said **** it, and ordered up 2 new GTX 680's last week. Of course, before this, I did ensure to try with different drivers (rollbacks, stable, and beta drivers - ensuring not to install that ****ing HD Audio Driver).

New cards arrived, so I decided to throw the rest of the RAM in while I was at it. After this, the System failed POST, showing up Q-Code 69 (System Agent DXE initialization is started). Now, the odd thing is that when this happened, for the first time ever, the Cyclone fan mounted on the RAM actually spun up to 4200rpm, and remained stable there.

Process of elimination started, and I found the system would only pass POST if the RAM Sticks were not connected.

Finally, into Windows - installed the newest drivers from nvidia.com (I of course removed the old drivers and used driversweep to clear out any remnants), and the crashes started; very constantly. The PC would normally completely lock during/just after the windows logon screen. At times, the PC wouldn't completely lock, but would freeze up for 5 seconds, respond for a second, freeze up again.. etc etc, leading to a hard reboot. I started the process of again trying every driver available, to no avail. At one point, I got an actual error message out of it; "Display driver nvlddmkm stopped responding and has successfully recovered.".

So, after a restless night of sleep, I woke up today and went back at it - and found I'm now back to the same old "stability" when I'm running on a standalone GTX680.

So, the question is - what exactly is ****ed, or are there any further tests I can do on this?

deconduo

Well first thing to do is to run memtest to see which sticks of RAM are faulty and RMA them. That might be the only problem, but it doesn't sound like it.

Another problem could just be bad overclocking, just because it boots doesn't mean it stable. You might need to push up the voltage to keep it stable, or it could just be an unlucky chip that caps out at 4.6 or something.

Apart from that, because it seems to be an intermittent problem its harder to pinpoint. However next step would be to breadboard your system and run some stability benchmarks like furmark and prime95.

If you're based in Dublin I could pop over to help you out if needed.

Tellox

deconduo wrote: »

Well first thing to do is to run memtest to see which sticks of RAM are faulty and RMA them. That might be the only problem, but it doesn't sound like it.

Another problem could just be bad overclocking, just because it boots doesn't mean it stable. You might need to push up the voltage to keep it stable, or it could just be an unlucky chip that caps out at 4.6 or something.

Apart from that, because it seems to be an intermittent problem its harder to pinpoint. However next step would be to breadboard your system and run some stability benchmarks like furmark and prime95.

If you're based in Dublin I could pop over to help you out if needed.

Sorry, should've mentioned - I ran a 3-pass test on the RAM itself, all came out good and clear. The cpu isn't overclocked anymore, it's back to its stock 3.2ghz.

Unfortunately, I'm based in Cork

deconduo

Tellox wrote: »

Sorry, should've mentioned - I ran a 3-pass test on the RAM itself, all came out good and clear. The cpu isn't overclocked anymore, it's back to its stock 3.2ghz.

Unfortunately, I'm based in Cork

So did all the RAM sticks pass?

I'd try breadboarding it now then. Take it out of the case and put it on a block of wood. Just put the CPU, one stick of RAM and one of the 680s in. Connect up one hard drive and do a fresh format + windows install. Run stress tests for a few hours to see if the problem occurs. If it does, switch out one component at a time to see if you can narrow down the problematic part.

Tea_Bag

just to add, run Intel burn test instead of prime95.

was doing some overclocking all day yesterday, and ran prime95 (because its easier to tell if a single core fails or catastrophic failure), but I ran it for hours on end, and had no problems. put on 30 passes of Intel burn test and it failed after 20min.

general consensus on the Internets is IBT is more stressful than P95 this gen.