Did I damage my CPU?

[dinkumthinkum]

I have a server (no overclocking), a few years old, with an E8400. I just had to move it from one place to another and now for the past day and a half I find it shutting itself down randomly. I do some diagnostics, there's machine check exceptions regarding thermal events, so I install lm-sensors and check the temps. It's idling in the high-70s, mid-80s. Under load it's spiking to 100C. I reboot and open the BIOS health status screen to be sure that I am reading the right temps. The moment I open it, the CPU temp is 100C, and climbing fast. Before I can even hit the power button, it is at 112C and then shuts itself down.

I am about to go ahead and reseat the HSF with a fresh coating of Arctic Silver. But should I just go ahead and replace the CPU since it is already 3 years old and I may have cut its lifespan terribly?

IDLE:

it8718-isa-0290
Adapter: ISA adapter
in0:          +1.06 V  (min =  +0.00 V, max =  +4.08 V)
in1:          +2.03 V  (min =  +0.00 V, max =  +4.08 V)
in2:          +3.30 V  (min =  +0.00 V, max =  +4.08 V)
+5V:          +2.85 V  (min =  +0.00 V, max =  +4.08 V)
in4:          +4.08 V  (min =  +0.00 V, max =  +4.08 V)  ALARM
in5:          +0.05 V  (min =  +0.00 V, max =  +4.08 V)
in6:          +4.08 V  (min =  +0.00 V, max =  +4.08 V)  ALARM
in7:          +3.01 V  (min =  +0.00 V, max =  +4.08 V)
Vbat:         +3.07 V  
fan1:        1717 RPM  (min =    0 RPM)
fan2:           0 RPM  (min =    0 RPM)
temp1:        +40.0°C  (low  = +127.0°C, high = +127.0°C)  sensor = thermistor
temp2:        +67.0°C  (low  = +127.0°C, high = +127.0°C)  sensor = thermal diode
temp3:         -2.0°C  (low  = +127.0°C, high = +127.0°C)  sensor = thermistor

coretemp-isa-0000
Adapter: ISA adapter
Core 0:       +78.0°C  (high = +78.0°C, crit = +100.0°C)  ALARM (CRIT)

coretemp-isa-0001
Adapter: ISA adapter
Core 1:       +77.0°C  (high = +78.0°C, crit = +100.0°C)  ALARM (CRIT)


LOAD:

it8718-isa-0290
Adapter: ISA adapter
in0:          +1.09 V  (min =  +0.00 V, max =  +4.08 V)
in1:          +2.03 V  (min =  +0.00 V, max =  +4.08 V)
in2:          +3.28 V  (min =  +0.00 V, max =  +4.08 V)
+5V:          +2.85 V  (min =  +0.00 V, max =  +4.08 V)
in4:          +4.08 V  (min =  +0.00 V, max =  +4.08 V)  ALARM
in5:          +0.14 V  (min =  +0.00 V, max =  +4.08 V)
in6:          +4.08 V  (min =  +0.00 V, max =  +4.08 V)  ALARM
in7:          +3.01 V  (min =  +0.00 V, max =  +4.08 V)
Vbat:         +3.07 V  
fan1:        1708 RPM  (min =    0 RPM)
fan2:           0 RPM  (min =    0 RPM)
temp1:        +39.0°C  (low  = +127.0°C, high = +127.0°C)  sensor = thermistor
temp2:        +88.0°C  (low  = +127.0°C, high = +127.0°C)  sensor = thermal diode
temp3:         -2.0°C  (low  = +127.0°C, high = +127.0°C)  sensor = thermistor

coretemp-isa-0000
Adapter: ISA adapter
Core 0:      +100.0°C  (high = +78.0°C, crit = +100.0°C)  ALARM (CRIT)

coretemp-isa-0001
Adapter: ISA adapter
Core 1:       +98.0°C  (high = +78.0°C, crit = +100.0°C)  ALARM (CRIT)


LOG:

kernel: [ 1499.816014] [Hardware Error]: Machine check events logged
mcelog: HARDWARE ERROR. This is *NOT* a software problem!
mcelog: Please contact your hardware vendor
mcelog: MCE 0
mcelog: CPU 0 THERMAL EVENT TSC 373bbee5e1a 
mcelog: TIME 1307127492 Fri Jun  3 14:58:12 2011
mcelog: Processor 0 heated above trip temperature. Throttling enabled.
mcelog: Please check your system cooling. Performance will be impacted
mcelog: STATUS 88010023 MCGSTATUS 0
mcelog: MCGCAP 806 APICID 0 SOCKETID 0 
mcelog: CPUID Vendor Intel Family 6 Model 23
mcelog: HARDWARE ERROR. This is *NOT* a software problem!
mcelog: Please contact your hardware vendor
mcelog: MCE 1
mcelog: CPU 0 THERMAL EVENT TSC 373bc002b4d 
mcelog: TIME 1307127492 Fri Jun  3 14:58:12 2011
mcelog: Processor 0 below trip temperature. Throttling disabled
mcelog: STATUS 88010022 MCGSTATUS 0
mcelog: MCGCAP 806 APICID 0 SOCKETID 0 
mcelog: CPUID Vendor Intel Family 6 Model 23
kernel: [ 1616.571112] i2c /dev entries driver
kernel: [ 1752.243219] CPU0: Core temperature above threshold, cpu clock throttled (total events = 93581)
kernel: [ 1752.243609] CPU0: Core temperature/speed normal

 

[alanwest09872]

if it serves your purpose why replace it. Im a newb thou. New cpus are only as good as the software you want to use. If you use 3yr old software then it should be adequate. But again I am a newb.

 

[Phynaz]

Sounds like something didn't take the trip too well. I'd do a complete re-install of the CPU and heatsink.

 

[dinkumthinkum]

I am examining and cleaning the CPU right now. No external signs of damage. The HSF seemed to be slightly loose when I first looked at it, that may be the culprit. I am about to reset it with a fresh coat of thermal compound. But having hit temps of 112C, do you think that the CPU is likely to fail soon even if temps are restored to normal?

 

[Zap]

It will probably be fine. Is EIST enabled in BIOS as well as any protections?

FWIW I had an overclocked Pentium III that ran for a while with no heatsink on it whatsoever. The machine had gotten knocked over and the heatsink fell off the CPU - it was a socket 370 CPU in a slotket. I didn't know, and let it sit like that for a half hour. Came back to it and it was locked up. It was still fine after that. Intel CPUs are surprisingly robust with temperatures, especially at stock clock/voltages.

 

[VirtualLarry]

The CPU is still likely to be fine. If it was shutting down due to temps, that means that the thermal throttle/shutdown feature is indeed working fine, which should save the CPU from damage.

I would (carefully) replace the heatsink, possibly with a new one if the push-pins are damaged, and put new thermal paste on.

Overall, though, I think it should be alright.

 

[dinkumthinkum]

Well, the fun continues.

The heatsink fan isn't in the greatest shape, so I may have to replace it whatever may come. However the temps are back down to normal: 45 idle, 65 load. The problem now is that suddenly, without any warning, the machine freezes after a few minutes of use.

I tried a number of things. After the first couple times, I rebooted into single user mode and tried to examine the temperature sensors and the system log. The temps were right -- 45C, and there were no errors in the system log. However the machine froze on me while I was reading the log. This is not heavy duty, it's just a text viewer in the system console.

I booted into BIOS and read the PC Health status screen: CPU Temp = 37C. I let it run a few minutes and sure enough: it froze in the BIOS.

At this point, I think I have to conclude that the CPU is damaged. It could also be the memory, though I am not sure why running the BIOS for a few minutes would instigate a memory problem. What do you think?

 

[Phynaz]

Reseat everything at this point. Memory, cards, all cables etc.

 

[pm]

Yeah, I'm with phynaz - take everything out and put it all back in again. And I'd run a memory check.

 

[dinkumthinkum]

Status update: I went back to reseat everything and noticed that even after 5 minutes of being shut off, the chipset heatsink was pretty hot to the touch. Anyway, I made sure to re-assemble the full case so the airflow was properly restored. So far, it has not crashed. Tomorrow I will run some stress tests and memtest to be sure.

Thanks.

 

[Soulkeeper]

I assume you are using linux ?
I had machine check exceptions on my server too, although it was really 100% stable
for me I just disabled the checks in the kernel

Another option is updating to the latest kernel available for your distribution.

things i'd do:
reseat heatsinks
clean out all dust with compressor, psu too
check for loose cables
run case for a short time with side off to ensure all fans are spinning
then run memtest86+ for an hour or so


EDIT: sorry I didn't read all the replies above, you can disregard most of this

 

[dinkumthinkum]

I did recently update the kernel, 2.6.38-2-amd64 (Debian). However since the freezes occurred in the BIOS I assumed that it had nothing to do with the OS. The freezes did only occur with the case side off, so I started wondering if it had something to do with inadequate case cooling. I did notice that the Seasonic PSU fan was not spinning, any time I checked. I presumed that was because the PSU was doing fine, but maybe that is also a sign of something.