New Q6600 G0 Build Failing at Stock Speeds in Prime95

This sounds like a processor issue but I'm not sure. I just built this system yesterday and it seems to be stable but when I run prime95 it will cause my system to crash in both XP and x64 Vista. I have tried the different types of stress tests running 4 processes but it will usually crash in 10-30 minutes. Temps don't seem that high even though it's really hot in my room. I can barely detect any warmth when touching my Tuniq Tower and the only parts that seem hot were the 8800GTX which was idling at about 58c and the ram. In core temp which I can't get to work much none of the cores were over than 51c when it crashed in Vista.

I tried running the latest memtest for 6 hours and it didn't get any errors. I tried bumping the processor and ddr2 voltage up a little but it made no difference. The only other thing I could think of is after I was running memtest for a while it looks like everything was powered off and my UPS went out so maybe the system is not getting good power from the UPS or the outlet it's plugged into. A few days ago the UPS reported outages in my other computer that it recovered from so I'm going to try another outlet next.

Any suggestions?

Edit: It has been running for about an hour using the test with the small FFT and it probably hasn't crashed yet because it's a less stressful test. I will try to add a picture of core temp in a little.

http://img221.imageshack.us/img221/3079/63006124em7.jpg
Shot at 2007-08-01

The small FFT test ran for over 2 1/2 hours and and it would get the restarts on the blend and large FFT tests. Here is a screen shot the blend test which will probably fail but it has been running so far for 20 minutes so I'll see how long it lasts.

http://img529.imageshack.us/img529/6036/untitledqp1.jpg

Could you show the temps tab of uGuru?

Yeah, here it is with Prime95 running for over 40 minutes so far which is longer than the blend test lasted before.

http://img294.imageshack.us/img294/5...titled2eq5.jpg

I also just checked the NB HS and it wasn't that hot.

stock voltage is 1.3 isnt it? on core temp VID

OK, so the IP35 doesn't use digital PWM. That's a good thing.

I don't really know what the problem could be. Maybe increase the MCH voltage a bit.

The stock voltage was set to 1.3. It just froze after an hour on the blend test and all of the core temps were stable.

it shows 1.24 in uguru either its reading wrong or is drooping which you need to compensate for and see what happens after that.

Ok, I tried bumping up the CPU and MCH voltage. I'll see what happens with the stress test. What component could be causing this problem?

Now that I bumped up a few voltages guru shows that the CPU is at 1.29-1.30. Could an option like EIST in BIOS be causing the problem?

You properly setup the ram ? 4 gigs sometimes require a bit more juice to the NB... also dunno if your board has a drop in voltage when under load; if so put 1.32 or something to keep it at 1.3 when under load ( is that what the vid is when detected by coretemp ?)

I left the ram at automatic settings which is 5-5-5-15. I think it is dropping the voltage because when it was set to 1.3 which is the vid it would stay at 1.24 under load in uguru. I think I bumped it up to 1.37 and it shows that it's staying at 1.28 under load but it did increase my temps by about 3 degrees. The temps are up to 57c now which is not good for stock voltage. I also bumped the MCH up a little. I looked through the overclocking thread on this mobo and from the screen shots I didn't see that anyone had as low of a CPU voltage as I had in uguru.

So far the most stressful large FFT test has been running for an hour since I turned up voltages which is longer than it made it before so I'll have to see if that helped.

Prime95 has been running on the large FFT test for over 2 hours since I upped the CPU and MCH. I'll see if it can go to 4-5 hours. I was hoping to OC to 3-3.2 but now I'm not sure if I'll be able to do that with stable temps.

Vcore is set to 1.34V and it's running at 1.27V under load.

MCH 1.25V is set to 1.33 and is running at 1.32.

damn those abits really droop alot on the quads. you can try enabling C1E in the bios, it downvolts your cpu at idle so you bump vcore a little more to compensate for the vdroop.

I made sure that C1E is on and when I tried a stress test with it off it didn't stop the crashes. Others with the IP35 Pro seemed to be getting higher voltages during load so I'm wondering if I have to RMA the CPU or mobo. Right now the highest temps are staying at 56-57 with Prime95 running.

i had a bit of trouble with the board initially until i realised that it was ram related and i had to load bios failsafe defaults and enable some of the peripherals manually to get the mobo to be stable :up:

I'm not sure about my ram it seems to be running hot but it didn't appear to get any errors after 6 hours of memtest and I did clear bios a few times yesterday.

try what i suggested if a problem returns i guess what i'm saying

ok I'll try that as soon as this stress test is done

Firstly, have you disabled the logging feature of CoreTemp? That can cause stability issues.


dinos, does the abit P35 have a voltage droop limiter option like the ASUS' do? The vcore seems to be drooping down to ~1.24v which should be plenty for 2.4Ghz but......

nah it droops bad

it's got some issues with certain ram this board

I didn't disable logging in core temp but I wasn't running it a lot of the times it crashed.

Prime95 just failed after 3 1/2 hours after those voltages were upped so I'm messing around with options in bios now.

Maybe I'll try running it with 1 stick of ram.

The problem might have gone away when I downclocked the ram to 4:5 ddr2-640. I tried an 9x296 OC on the fsb and it wouldn't even get to bios on stock setting but when I downclocked the memory I was able to get into bios. Also, I'm running Prime95 with the the downclocked ram stock cpu settings without increasing voltage. If I hadn't reduced the ddr2 speed Prime95 probably would have already failed.

I'll have to try one stick at a time. Does this suggest that the memory could be faulty because memtest did pass?

Prime95 has run for 5 hours with stock fsb and voltage with the ddr2 underclocked and it hasn't frozen. So it looks like turning down the ram speed fixed the problem but I don't want to keep it running much slower than 800.

If underclocking the ram is the only way to get it running stable does that suggest faulty memory or something else?

could also be a bios limitation, but with 4 dimms heat might also cause instability, try dimm per dimm mate and see what gives and if possible give them to mate who can heck them out