Ran HWinfo for 30min with Prime95 (blend), couldn't get a crash again.
CPU temps are scary.
Hitting over 100º C.
Can't tell how the system did not freeze.
Well fixed Intel's BOX Cooler.
Should RMA the RAM tomorrow?
Would leave me without the machine for some time.
Maybe monday?
Core 1 started throttling at 104C and even shut down a number of times. The CPU fan was responding but couldn't get high enough rpm to keep the temps down and the "system-3" fan was lumbering along without doing much of anything. The CPU fan maxed out at 2566 from a low of 1844 and "system-3" fan maxed out at 1247 from a low of 1236. I can only guess that the system-3 fan is a chassis fan. You certainly need to move more air through the cabinet. I would check your BIOS settings and adjust fan speeds if your system allows it. Many default settings are too low for a high powered system.
In addition to carwiz's pristine incite, even with poor airflow within the chassis it still shouldn't increase in temp for the CPU that fast nor that high. I suspect the CPU cooler has been installed improperly, such as a bad thermal paste application or the heatsink is not sitting correctly. You'll want to remove it, clean everything up of thermal paste, and re-seat the cooler.
Though honestly, I have a bad feeling that your CPU (and perhaps even mobo) may have already been done in by the stress it's had to endure. Even if you happen to fix the cooling issue the symptoms may still persist, meaning the heat has taken its toll.
System 3 is a chassis fan, came along with the Carbide 500R.
Just took out the CPU cooling BOX and installed it again, just to be sure it was properly installed.
Stock cooler, thermal paste, heatsink..
Still getting the same temperatures at prime95, though CPU Fan speed increased to a maximum of 2733 RPM.
Maybe it's just that poor quality thermal paste that came along, I don't know..
Still waiting the next system crash.
So it finally crashed.
Weird it took so long...
Attaching data.
HWinfo was running, the .csv goes inside the folder also.
Thanks!
FAULTING_IP:
nvlddmkm+bd5d8
fffff880`049605d8 488b9200020000 mov rdx,qword ptr [rdx+200h]
EXCEPTION_RECORD: fffff88003bfd0b8 -- (.exr 0xfffff88003bfd0b8)
ExceptionAddress: fffff880049605d8 (nvlddmkm+0x00000000000bd5d8)
ExceptionCode: c0000005 (Access violation)
ExceptionFlags: 00000000
NumberParameters: 2
Parameter[0]: 0000000000000000
Parameter[1]: ffffffffffffffff
Attempt to read from address ffffffffffffffff
At this point it's hard it's hard to put a finger on a single fault except the heat. I don't know if it's a side effect of Prime or not at this point. Windows entered a ACPI Kernel Power fault state because of several more throttling events but this time on Core 0. It could be that Prime happened to be running on that core this time. I don't know if the invalid memory access (Nvlddmkm) is occurring because of faulty drivers or the hardware is breaking down due to heat. I suspect the heat since most addresses, invalid or not, will have something in them rather than all "f" or 0. It's obvious in the logs that the processor is shutting off the core for periods before the crash. This is a normal function of the processor during high heat.
I respectfully disagree with prior statements about not needing case cooling air movement. It does little good to recirculate hot air in the cabinet. That's like blocking the radiator of a car and expecting it to get cool air from the engine compartment. The Processor hit 86+ watts on numerous occasions during Prime and, no doubt, the reason for the throttling temperatures. The purpose of the case fan is to draw cooler air into the case from vents and exhausting heated air from the case. It's pretty obvious that the case fan is not controlled and runs at a fairly constant speed. Either the BIOS fan control parameters are incorrect or it's a single speed fan that can't keep up.
I don't know about temperature this time.
I was only surfing the internet using Waterfox.
Checked temperatures few minutes after the crash, it was on a constant variation from 45º to 50º C.
Prime95 was not running this time.
The very first BSODs that happened to the system were on idle, and had never stressed the CPU at that point.
This specific error had never happened before.
Whatever, my mind is all messed up with this computer.
Thanks for the help guys.
Just had another BSOD.
Data attached.
Thanks
The last two crashdumps are rather inconclusive. We'll need Driver Verifier on in order for the crashdumps to be worthwhile.
Got it.
DV was on.
HWinfo was logging, .csv goes with the folder.
Playing BF : BC2
Ran lot of Furmark' stress test earlier, all ok.
Data attached.
Last edited by rslembo; 15 Feb 2013 at 17:59.
Quote