CPU Voltage Strangeness...

[Draex]

Currently working on a E5200 on Asus P5E-VM that had some problems, the Thermal Compound had run between the base of the HSF/CPU and the machine had been left in operation (Shutdown on Temp had been disabled in the bios) the chip was running pretty consitantly at about 90-100C when I went to look at it, and the system was unstable to all getup, I replaced the HSF and the chip is now running at about 30C Idle 38 Load, but the instability remains (Crashes Prime95/Intel Burntest within Seconds) I have disabled speedstep and increased the voltage for various things (CPU voltage, FSB Foltage, PLL voltage ect) and Intel Burn Test will pass, but Prime95 will fail within 5 minutes, the otherthing I have noticed is CPU-Z reports the voltage as 1.168 volts (At load or idle) regardless of the voltage set in Bios, (Set at 1.225 vcore currently)

Is this chip done for, or is there something more I can do dredge some more life out of it?

Thanks for any input.

[0o0]

just out of curiosity, why did you have throttling and shutdown disabled? I'd say it's toast, no way to fix a cpu like that. I'm still not quite convinced though.. you say the the maximium temperature was 100 and not any higher?

[Draex]

Its a friends system not my own, so I cant speak for why thermal shutdown was turned off, I disabled C1E (Speedstep) as it caused the system to crash as soon as it tried to change the multiplier.

That said I have made abit more progress, I removed one of 2 Dimm's, it does not seam to matter which as they both pass tests in Memtest in another machine, but if 2 Dimm's are in I get the crashing in Prime95, after droping it to one Dimm, and increasing the dimm voltage to 2.0, and relaxing the timings to 6-6-6-15 (the sticks are labeled for 1.8) I've been prime95 stable for a few hours now.

Still cant for the life of me figure out why the vcore voltage wont change regardless of setting in the bios.

And I don't believe the temperature went that much higher then about 100C, it probably hovered there for a month or more though.

[clone63]

The symptoms described would make me assume motherboard. BUT considering the CPU's handling, I wouldn't be surprised if it was now toast. Do you have another cpu to test on the board?

[rjbarker]

Did you try and "clean install" of the OS as well as Clear CMOS after re-installing the CPU?

Even if you didn't physically pull he CPU from the Board (just clean and install new TIM) I would still try this prior to ditching the CPU.

[Draex]

I haven't tried another CPU in the board yet but I do have some that I could use if needed, I did physically remove the CPU when cleaning it from its socket, I did not see any discoloration on the bottom of it.

The system is stable as of this morning with 1 Dimm installed, but the CPU voltage is still not responding to adjustments, and it still fails with any 2 Dimm's installed regardless of what banks they are placed in.

[geokilla]

Reset everything back to BIOS defaults. Don't use Prime95, use OCCT instead. Report back. Weird to see how it can pass Linpack (IBT) but fail Prime95...

Also as you said, run just 1 RAM stick. Could be bad RAM...

[rjbarker]

Clear CMOS

[gingerbee]

um this sounds MB to me i have had friedns chips run much hotter with no fan on it and they still run fine now after 6 7 years so i would try another cpu first to see if it the MB. jm2cents
Good luck

[myleftnut]

I've had an E5200 run at ~100C before for quite some time and it was fine so I highly doubt it's the cpu. Sounds like there might be a few things causing problems: RAM and Motherboard. Since it doesn't sound like the motherboard was stressed with high volts and high work loads, I'm more inclined to believe it's the ram. Although, northbridge cooling could also be an issue since this board does have onboard video. Have you tried sticking a fan to the NB, turning off onboard video and using a dedicated videocard?